Category: SeaPRwire

17.5 against 40.7 Start with the ratio, because it is the one that does not get quoted. – AsiaPresswire – In the first quarter of 2025, non-oil and gas manufacturing accounted for 17.50% of Indonesian GDP — Rp 991,785.2 billion of output at current prices, according to Statistics Indonesia. In the twelve months to the second quarter of 2025, the same sector absorbed 40.72% of total realised national investment: Rp 568.4 trillion, according to Industry Minister Agus Gumiwang Kartasasmita’s October 2025 press conference. Divide one by the other. Manufacturing takes about 2.33 times its share of GDP in investment. That is the whole argument, in one ratio. A sector producing roughly a sixth of output is consuming roughly two fifths of the capital being deployed in the country. Either that is a deliberate bet on industrialisation that will pay off later — which is the government’s position, and a defensible one — or it is capital being spent at a poor rate of conversion. Both readings are compatible with the data available. The data available does not, on its own, settle which is true. The third number in the set is jobs. That same Rp 568.4 trillion of manufacturing investment is credited with 19.44 million workers absorbed, 13.41% of national employment. So: 40.7% of the capital, 17.5% of the output, 13.4% of the jobs. Read in that order, the sector is capital-hungry relative to what it currently produces and to whom it currently employs. That is what a country in the middle of a downstreaming push looks like. It is also what a country with an expensive logistics base looks like. The rest of this piece is about telling those two apart. Eight months of PMI, read in order The monthly sentiment data is where the volatility lives, and reading it as a series rather than as a headline is the entire trick. S&P Global’s Indonesia Manufacturing PMI, 2025: January: 51.9 — moderate expansion, the highest reading since May 2024. February: 53.6 — strongest growth in eleven months, driven by new orders and output. March: 52.4 — slower than February, still comfortably above 50. April: 46.7 — sharp contraction, the steepest since August 2021. May: 47.4 — still contracting, slightly less badly. August: 51.5 — back above the line. September: 50.4 — slower growth. October: 51.2 — third consecutive monthly improvement, with employment rising at the steepest pace in five months. The range across eight months is 6.9 points. That is not a sector gently trending. That is a sector whose monthly readings swing between “strongest in nearly a year” and “worst in nearly four years” inside ten weeks. Two things sit behind the April collapse. Domestic demand weakened after the post-Eid consumption period, and export orders to the United States fell as tariff uncertainty hit. Layoffs in the wider economy passed 24,000 workers by May 2025, which feeds straight back into domestic demand for manufactured goods. Input costs rose at the same time, and manufacturers absorbed most of it: factory gate prices rose only fractionally, because raising them would have cost volume. October’s reading is the more informative one, because of what did not happen. New orders rose for a third month at a pace equal to the joint-strongest since March. Employment rose. But production was flat, and panel members attributed that partly to having run down finished-goods inventory in prior months. Export orders fell for the second consecutive month. Usamah Bhatti, economist at S&P Global Market Intelligence, put it this way in the October release: “The improvement in the health of the Indonesian manufacturing sector strengthened at the start of the fourth quarter of 2025, boding well for the coming months.” He also flagged the price side — “Firms were hesitant to fully pass through higher expenses to clients, as factory gate charges rose only fractionally as part of efforts to remain competitive on price.” The quarter-by-quarter share series The GDP share figures are more stable than the PMI and more useful than any single annual number, because they show the direction. Non-oil and gas manufacturing as a share of GDP at current prices, by quarter: Q1 2024: 17.47% (Rp 923,722.5 bn) Q2 2024: 16.70% (Rp 924,737.8 bn) Q3 2024: 17.18% (Rp 968,780.8 bn) Q4 2024: 17.31% (Rp 982,048.2 bn) Q1 2025: 17.50% (Rp 991,785.2 bn) The spread is 0.8 percentage points across five quarters. The sector is not collapsing and it is not surging; it is holding a share of roughly 17% to 17.5% while the rest of the economy grows around it. There is a longer arc behind those five quarters, and it points the other way. Annual estimates put manufacturing’s share near 19.3% in 2020, then 18.9%, 18.6%, 18.7%, around 18.5% in 2024, and a projected 18.2% for 2025. On that measure, Indonesian manufacturing has been losing share for five years. The two series do not contradict each other. The quarterly BPS figures measure non-oil and gas manufacturing; the annual series is a broader definition. But the direction they jointly suggest is that Indonesia is holding its factory base roughly flat as a share of the economy, against a trend of slow erosion that predates the current administration. Growth rates sit alongside this. The minister put sector growth at 4.94% from the fourth quarter of 2024 to the second quarter of 2025. BPS put non-oil and gas manufacturing growth at 4.31% year on year in Q1 2025, with food and beverage expanding 6.04% — the fastest large sub-sector, and the one that contributes over 30% of manufacturing GDP. Splitting Rp568.4 trillion by worker Per-unit arithmetic is where big national figures stop being impressive and start being checkable. Take Rp 568.4 trillion of annual manufacturing investment against 19.44 million workers in the sector. That is roughly Rp 29.2 million per employed person, or about USD 1,780 at an exchange rate in the region of 16,000 rupiah to the dollar. That figure needs two caveats before anyone uses it. The Rp 568.4 trillion is a flow — one year of realised investment. The 19.44 million is a stock — everyone currently employed in the sector. Dividing a flow by a stock gives a ratio, not a cost per job created, and the real cost per *new* job is several times higher. The rupiah figure is also sensitive to the exchange rate, and 2025 was not a stable year for it. With those caveats, the ratio still tells you something: Indonesia is deploying capital into manufacturing at a rate of roughly Rp 29 million per person already working there. Compare that with a sector where the median manufacturing wage is a small fraction of that figure, and the gap is the capital intensity of downstreaming. Smelters, refining and chemical processing are not labour-intensive. They are capital-intensive by design, and Indonesia chose them deliberately. Per-worker output tells the same story from the other side. Quarterly manufacturing output of roughly Rp 991,785 billion across 19.44 million workers is about Rp 51 million per worker per quarter, or roughly Rp 204 million a year — around USD 12,700 of gross output per person. Again: gross output, not value added, and not a wage. But it is the number to hold when someone tells you Indonesian manufacturing is low-productivity, and it is the number that makes the logistics figure below hurt. 25% exported, 75% sold at home The minister made a point in October 2025 that deserves more attention than the growth number: Indonesian manufactured goods go 25% to export markets and 75% to the domestic market. This is the structural fact that separates Indonesia from its neighbours. Vietnam and Thailand run export-led manufacturing sectors. Indonesia runs a domestically-absorbed one. The trade data looks weak by comparison — the minister acknowledged Indonesian exports trail Malaysia, Vietnam and Thailand — and the explanation offered was that this is a market-orientation choice rather than a competitiveness failure. Both things can be true. A large domestic market is a genuine advantage: 280 million people absorb a great deal of what the country makes, and that demand does not vanish when a foreign tariff changes. It also means the sector is exposed to domestic purchasing power in a way Vietnam’s is not, which is precisely what the April 2025 PMI collapse demonstrated. When domestic demand weakened, factories cut output within weeks. The export share is where the policy energy is going. Hilirisasi — downstream processing — is explicitly designed to move Indonesia up the export value chain: not nickel ore but nickel products, not bauxite but aluminium. Central Sulawesi and North Maluku have become investment hotspots on exactly this logic. Food and beverage: over 30% of manufacturing GDP, growing 6.04% in Q1 2025, overwhelmingly domestic. Automotive: driven by domestic demand and EV-related investment, concentrated in West Java. Chemicals and pharmaceuticals: expanding on health demand and industrial inputs. Textiles and apparel: facing export competition, pivoting toward sustainable fibres. Electronics: still modest, growing on regional supply chain shifts. Only two of those five are meaningfully export-oriented today. 23.5% of GDP spent on moving things Here is the number that explains the most and gets discussed the least. Indonesia’s logistics cost is estimated at 23.5% of GDP, against 16.8% for Vietnam and 14.5% for Thailand, on World Bank figures. That is a gap of roughly 6.7 percentage points against Vietnam and 9 points against Thailand. Put it against the sector it bites hardest. If manufacturing is about 17.5% of GDP, then a logistics burden nine points above a direct competitor is a tax on every factory that has to move anything — which is all of them. On a per-unit basis it is the difference between a West Java plant being competitive for an export order and not being competitive, before a single wage or tariff is considered. Why is it so high? The answers are structural and unglamorous: An archipelago with a road network concentrated on Java and inter-island movement dependent on shipping. Port dwell times and handling costs that are improving but remain above regional comparators. A trucking sector with low average fleet age and high empty-running rates. Warehousing concentrated around Jakarta, so anything made elsewhere travels to Java before it travels anywhere else. That last point interacts with the next section. Manufacturing is concentrated in Java partly for good reasons — infrastructure, market proximity, suppliers — and that concentration then generates logistics costs for everyone outside Java who wants to participate in the same supply chains. 170 parks, 95,000 hectares, 59.5% full As of May 2025, 170 industrial park operators held licences, covering close to 95,000 hectares, with an occupancy rate of 59.52%. Sixty per cent occupancy is the number to sit with. It means roughly 38,000 hectares of serviced, licensed, industrial-zoned land is currently empty. That is not necessarily a bad thing — a park needs vacancy to absorb a large new tenant, and 40% vacancy is not unusual for a young estate — but it does complicate the argument that Indonesia’s constraint is a shortage of industrial land. If 40% of licensed park land is empty, the binding constraint is somewhere else: power, water, port access, or the cost of getting a tenant through licensing fast enough to matter. The Ministry of Industry’s response has been to draft a regulation for Special Industrial Zones, or KIT, intended for smaller or themed parks in land-constrained or strategically located areas. That is a sensible instrument for the land-constrained case. It does nothing for the empty-38,000-hectares case. The occupancy figure also hides enormous variation. A park in Bekasi or Karawang near Jakarta will be near full. A park in Central Sulawesi attached to a nickel project will be full by definition. A park in a secondary province without a dedicated power plant or a port agreement will sit at a fraction of that, and will sit there for years. Six months in Java, eighteen in Kalimantan The regional distribution is the most concentrated part of the whole picture. Java hosts more than 75% of Indonesia’s medium- and large-scale manufacturing establishments. West Java alone accounts for over a third of the national total, by virtue of sitting next to Jakarta with access to ports and a dense supplier base. Sumatra and Sulawesi are growing, mostly on resource processing. Kalimantan is growing from a very low base. Then there is the timeline problem. Land acquisition for industrial use takes roughly six months in Java. In Kalimantan or Papua, it can run beyond eighteen months, driven by conflicting local land-use rules and overlapping claims. An eighteen-month differential is not an inconvenience. It is the difference between catching an investment cycle and missing it. A company evaluating two sites will discount the second one by the carry cost of eighteen months of capital, plus the risk that the land is still not clear at the end of it. That is how a country ends up with 75% of its factories on one island while officially pursuing regional equity. Layer on the licensing picture. The Omnibus Law was written to simplify exactly this, but reporting cited by industry analysts indicated fewer than 40% of provinces had fully implemented its provisions by 2023, per Ministry of Law and Human Rights data. A national reform that lands in under half the provinces produces a country where the answer to “how long will this take” depends entirely on which map you are standing on. Where the capital came from Manufacturing foreign direct investment in the first quarter of 2025 totalled USD 28.2 billion, up about 14% year on year. The source list is a map of the region’s capital flows: Singapore, China, Hong Kong, South Korea and Japan. Destination hotspots by commodity: Central Sulawesi for nickel, North Maluku for smelters, West Java for automotive and electronics. Twenty-eight point two billion dollars in a quarter is a large number, and it is worth noting what it mostly is: capital for processing facilities, not for light manufacturing employment. A nickel smelter in Central Sulawesi might employ a few thousand people for a billion dollars of investment. A garment factory in West Java might employ the same number for a few tens of millions. That is not a criticism of downstreaming. It is an explanation of the 40.7%-investment / 13.4%-employment pair at the top of this piece. Indonesia is choosing capital-intensive processing because it has the ore and it wants the value added, and it is accepting a lower employment yield per dollar as a result. Two policy levers are currently in play. The Domestic Component Level, or TKDN, is being relaxed, with President Prabowo Subianto asking his administration to ease requirements to attract more foreign investment. And the KIT regulation is being drafted to allow smaller, themed parks. Both push in the same direction: make it easier for foreign capital to land. Neither addresses the logistics cost. What 6.9 points of swing costs a factory The PMI is a diffusion index, not a volume index, and it is worth translating what a 6.9-point range actually does to a factory floor. A reading of 53.6 in February and 46.7 in April is not “slightly better” and “slightly worse.” Above 50, more firms report improvement than deterioration; below it, the reverse. April’s 46.7 was reported as the steepest contraction since August 2021, which means a majority of panel firms reported conditions deteriorating, and deteriorating at a rate not seen in nearly four years. For a factory manager, that translates into three concrete costs. A sector that swings this fast has to hold more buffer stock than one that trends. Buffer stock is working capital, and working capital in Indonesia is not cheap. Output flat while employment rises — the October pattern — means labour is being held through a soft patch rather than shed. That is good for the workers and expensive for the balance sheet, and it is visible in the October data as the steepest employment rise in five months alongside unchanged production. Input costs rose at the strongest rate in eight months in October while factory gate prices rose only fractionally. Margin compression of that kind does not show up in GDP. It shows up in the next investment decision. The range also matters for anyone modelling the sector. A single point estimate for Indonesian manufacturing growth in 2025 is close to meaningless when the monthly sentiment data spans 46.7 to 53.6 inside one year. Any credible forecast has to be a range, and the range has to be wide — roughly four to five percent growth if domestic demand holds, and something nearer flat if the post-Eid pattern of 2025 repeats and export orders to the United States stay soft. That is the honest position. The sector is expanding, the employment signal turned positive in October, and the swing is large enough that anyone claiming precision about the next two quarters is guessing. Two numbers to check next quarter The argument about Indonesian manufacturing will not be settled by the PMI, which will keep swinging, or by the GDP share, which moves in fractions of a point. Check these two instead. Occupancy rate of licensed industrial parks. It was 59.52% in May 2025. If it rises, the capital is landing and the downstreaming bet is converting into operating plants. If it stays flat while investment keeps rising, the money is going into announcements and land banks rather than production. This is the single cleanest indicator available and it is published. Manufacturing share of GDP at current prices. It was 17.50% in Q1 2025, up from 16.70% a year earlier. If the next four quarters hold above 17.5%, the sector is genuinely growing faster than the economy. If it drifts back toward 17% and then below, the 40.7% investment share is buying market share in nothing. Everything else — the logistics percentage, the land acquisition timelines, the TKDN negotiation — is the mechanism. These two are the readout.

Muhaisnah 4, before the shift – AsiaPresswire – The cabinet is about waist height, beige, and mostly unremarkable until you look at the plate. Then it says 360 kW, and it stops being furniture. Three hundred and sixty kilowatts is not a home wallbox. It is a piece of industrial equipment that happens to be standing in a staff car park. Dubai Taxi Company put the first two of these hubs where its own fleet sleeps: one at the depot near Dubai International Airport, one at the headquarters in Muhaisnah 4. Both are places where a taxi can be plugged in while its driver is doing something else — signing off a shift, drinking tea, waiting for the dispatcher to call a number. I have not driven a Dubai taxi shift, and I want to be plain about that before the scene gets going. What follows is assembled from Dubai Taxi Company’s 2025 annual and sustainability reporting, DEWA’s published charger and tariff material, the Roads and Transport Authority’s fleet history, and the access rules the utility publishes for its own network. The equipment, the numbers and the locations are documented. The shift described below is what that equipment makes possible, reconstructed rather than transcribed. The reconstruction is not complicated. A driver comes off the road with a car that has been working since before dawn. He plugs in. The cable is heavy and the connector seats with a sound like a truck door. He walks to the canteen. If the cabinet behaves as rated, he is back on the rank inside half an hour, which is the number that decides whether this whole programme works or does not. What 360 kW means on a working day The agreement behind these chargers was signed at WETEX 2025 between DEWA and Dubai Taxi Company, under DEWA’s “EV Green Charger” programme. It commits 208 ultra-fast charging points across the emirate, with individual units rated up to 360 kW, and a pathway to expand the network to 354 points by 2040. Two hubs open first — the airport depot and Muhaisnah 4. The rest of the 208 arrive over the following 12 to 18 months. The arithmetic of a taxi fleet is unforgiving, and it is worth doing out loud. A taxi that is charging is not earning. A taxi that charges slowly is a taxi that has to be scheduled around its own battery. A 360 kW unit, on a vehicle with a battery that will accept the rate, turns a charging stop into something closer to a fuel stop: a break of well under thirty minutes rather than an hour and a half. Multiply that by a shift. If a driver can take one short top-up mid-shift rather than losing the car for a long block, the fleet needs fewer vehicles to cover the same number of trips. That is the actual economic case, and it has nothing to do with emissions. Emissions are the reason the programme gets funded. Utilisation is the reason it gets built. DEWA and DTC estimate the first phase cuts roughly 37,939 tonnes of CO2 a year, rising to as much as 49,654 tonnes once the existing chargers are counted alongside the new ones. 380 Seals, and the range figure that matters The vehicles arrived before most of the chargers did, which is the normal order of these things and also the normal source of trouble. In 2025 Dubai Taxi Company partnered with Al-Futtaim Electric Mobility to put 380 all-electric BYD Seal sedans into service. DTC’s own sustainability reporting puts the figure of electric taxis deployed during the financial year at 435. The Seal’s stated range is up to 370 kilometres per charge, and the model was chosen over a hybrid for a reason: it is a full battery-electric car with enough range to cover a Dubai working day with air conditioning running, which is the only range test that means anything here. Three hundred and seventy kilometres sounds like a lot until you do the shift maths. A Dubai taxi working a full day with the compressor running covers a great deal of ground, sits in Sheikh Zayed Road traffic with the air conditioning at full load, and idles at ranks. Range in a hot climate is not a distance figure, it is a cooling figure. The 370 km number is a manufacturer figure, and the fleet number will be lower. DTC says 91% of its fleet is already hybrid or electric. That is a striking figure and it is worth reading carefully: hybrid is doing most of the work in it. Hybrid vehicles have been in the fleet for years, and they are the reason the company can make the claim honestly while still having a great deal of electrification left to do. Hybrid trials started in 2008 This is not a new programme, and the people running it know it is not. Dubai’s roads and transport authority began putting hybrid vehicles into taxi service in 2008, seventeen years before the DEWA agreement. The reported result was a 34% reduction in fuel consumption and emissions against the conventional vehicles they replaced. That is a large number achieved with a technology that nobody would now call ambitious. The sequencing matters. Hybrids first, because they need no infrastructure at all and because the fuel saving shows up immediately in the operating account. Then plug-in hybrids, then a partial battery-electric fleet on depot charging where the routes are predictable, and only then a large battery-electric fleet dependent on public ultra-fast infrastructure. That last step is the one Dubai is taking now, and it is the step where the infrastructure has to arrive on time. The 2008 trial did not require a single charger to be installed anywhere. The 2025 programme requires 208 of them to be installed on schedule, in a city where every site needs a grid connection, a landlord, and a wayleave. The tariff on the screen Charging is not free and never was. DEWA has said pricing for the new fleet points will follow its standard “EV Green Charger” tariffs, with the possibility of special rates for fleet operators. Access is through the DEWA app, smart cards, or contactless payment. Fleet operators register through the utility’s portal and manage drivers underneath a commercial account. For the driver, the interesting question is not what a kilowatt-hour costs. It is what the company does with the cost. If charging is deducted from a driver’s settlement, behaviour changes immediately and not always in useful ways — drivers skip top-ups, run low, and idle. If charging sits on the company account, the utilisation case works and the emissions case works, and the operator carries a new line in its cost base. The utility’s own network has been running since 2014 and already supports more than 1,500 charging points across Dubai. The DTC agreement is not the beginning of anything. It is the first time the network has been built around one fleet’s operational timetable rather than around general public access, and that is a different engineering problem: peak demand concentrated at shift-change, at two specific sites, every single day. Who absorbs the downtime The hard part of fleet electrification is not the cars. A taxi off the road for charging is a taxi not on the rank, and ranks are how Dubai’s system allocates work. Charging demand peaks exactly when the fleet wants to hand over: early morning and early evening. Two hubs at the start means two places where a queue forms, and a queue at a 360 kW point is a scheduling failure, not a technology failure. Battery-electric taxis have a different maintenance profile, which means the workshop has to learn something new at the same time as it is learning the chargers. DTC’s reporting shows it has been preparing for the maintenance side. It introduced an electronic incident reporting system during 2025, developed EV-specific business continuity and emergency response protocols, and tested what happens when the charging infrastructure is unavailable. That last item is a detail most fleet operators do not publish, and it is the one that tells you whether a company has thought about failure or only about rollout. The other absorbing mechanism is capital. Two hundred and eight ultra-fast points is not a small project, and it is being built by the utility rather than the fleet operator, which spreads the cost across DEWA’s rate base rather than putting it on DTC’s balance sheet. That arrangement is why the deal exists at all. What DTC said in its own report The company’s 2025 disclosures are unusually specific, and three lines are worth quoting rather than paraphrasing. Fleet electrification: 435 electric taxis deployed during the year, alongside the installation of 208 ultra-fast charging stations, described as expandable to 354 by 2040. Long-term target: full electrification of the taxi and limousine fleet by 2040, with the electric fleet expected to pass 2,500 vehicles by 2030. Health and safety: an electronic incident reporting system, structured training for drivers and staff supporting People of Determination, and a Driver Mental Health and Wellbeing Programme. That last cluster is easy to skip past in a sustainability report. It should not be skipped, because a fleet electrification programme is also a workforce change programme. Drivers who have spent fifteen years reading a fuel gauge are now managing a state of charge, an app, and a queue. DTC’s decision to put a mental health programme in the same report as the charger count is more informative than any of the emissions figures, because it means someone inside the company understood the shape of the change. The company also joined the British Safety Council during 2025 and built a digital ESG data collection tool, which is the sort of thing that reads like boilerplate and matters when a fleet is trying to prove to a regulator or a lender that its emissions claims are measured rather than asserted. Solar on the workshop roof, capped by regulation There is a quiet line in the energy section of DTC’s 2025 sustainability report that says more about the constraints than the achievements do. The company installed solar panels across its workshop and accommodation areas. Those panels supply approximately 25% of its energy needs — and the report describes that as the maximum solar capacity currently permitted under existing regulations. Read that again. A taxi company with large flat roofs in one of the sunniest cities on earth has hit a regulatory ceiling on how much of its own power it can generate. The constraint is not technical. It is grid interconnection rules written for a different era, and it is the kind of ceiling that will quietly cap how green an electrified fleet can actually become. The rest of the efficiency work is more ordinary and has already produced measurable numbers: 470,000 kWh saved annually from LED lighting, 3,050,000 kWh saved annually from more efficient air conditioning, and roughly half of the HVAC units across driver accommodation upgraded or adapted. In this climate, air conditioning is the load. Everything else is rounding. The driver mental health line in a sustainability report I want to stay with the mental health programme for a paragraph, because it is the least likely thing to appear in an article about electric vehicles and arguably the most diagnostic. A taxi fleet converting to electric changes the texture of a driver’s day. The car is quieter, which drivers generally like. The car needs planning, which drivers generally do not. Range anxiety is a real phenomenon for the first few months and then it becomes routine, but the period in between is exactly when a driver’s earnings are most at risk and when a company’s attrition is most likely. DTC reported a Driver Mental Health and Wellbeing Programme in the same year it deployed 435 electric vehicles and started a 208-point charging rollout. The sequencing is not accidental. A company that has run hybrid trials since 2008 has watched enough technology transitions to know that the failure mode is rarely the machine. The company also reported AED 3.54 million in approved community and charitable contributions during the year, along with transport support and in-kind services. That is a governance line, not an operational one, but it sits in the same document as the charging figures, which tells you something about how the company wants the whole programme read. Airport depot, and why two hubs first Putting the first two hubs on company property is the single most telling decision in the whole rollout. The airport depot and the Muhaisnah 4 headquarters have three things in common that a public site does not: DTC already owns or controls the land, the vehicles are already there overnight, and the grid connection is a negotiation with one landlord rather than with a municipality and a retail landlord and a facilities contractor. It is also a hedge. Depot charging is the easy case. The vehicles are parked for hours, the demand is predictable, and if a cabinet is down there is another one twenty metres away and a workshop next door that can look at it. A public ultra-fast hub on a rank or at a mall has none of those properties: the driver cannot wait, the dwell time is minutes rather than hours, and a fault means a taxi is off the road rather than late out of the depot. Every fleet electrification programme in the world starts on its own property for exactly these reasons. The interesting question is when it leaves. DTC’s 208 points are described as being deployed “across Dubai” rather than at its two sites, and the 12 to 18 month window is when that claim gets tested. Depot chargers can be installed quickly. Kerbside and rank-side chargers require wayleaves, distribution capacity at the specific feeder, and in most cases a civil works package. Watch the second and third tranches of sites rather than the first. If they are still on DTC land, the programme is a depot programme with a public-facing press release. If they start appearing at ranks, malls and transport nodes, it is what it says it is. Rank allocation, and the queue left unmodelled Dubai’s taxi system runs on ranks. A driver’s earnings are substantially a function of where he sits in a queue and how quickly the queue moves, and that makes queueing physics an industrial relations question rather than a traffic question. Electrification inserts a new constraint into that system. A driver who needs to charge cannot take the next fare. A driver who arrives at a rank with 12% state of charge is a driver whose next fare is a decision rather than a default. And a rank where three chargers serve forty taxis is a rank where the charging arrangement has to be allocated by somebody — a dispatcher, an algorithm, or seniority. Who gets the charger at shift change, and how is that decided? What happens to a driver who is dispatched a long airport run on a low battery? Is charging time paid time, unpaid time, or a deduction against the settlement? Does the dispatch system see state of charge at all, or only location? None of these questions appears in an infrastructure announcement, and all of them determine whether drivers experience this as an upgrade or as a tax on their day. DTC’s decision to publish a mental health programme in the same reporting year suggests the company has at least some of them on a whiteboard. For a fleet operator elsewhere in the Gulf watching this, that is the transferable lesson. The charger count is the easy part to copy. The dispatch integration is the part that decides whether the fleet actually runs. Limousines, and the half of the fleet left uncounted DTC’s electrification target covers taxis and limousines, and the second category behaves completely differently. A limousine working airport transfers and hotel runs has a duty cycle that is easier to electrify than a rank taxi. The trips are longer, the start and end points are known in advance, and there is dwell time at a terminal or a hotel portico where the driver is waiting anyway. It is the closest thing in the industry to a scheduled route, and scheduled routes are what battery-electric vehicles are genuinely good at. A rank taxi is the opposite. Unpredictable trips, unpredictable idle, air conditioning at full load in static traffic for long stretches, and no natural charging window other than the one the driver makes. The fleet also is not one vehicle type. Alongside the 380 BYD Seals, DTC’s reporting and trade coverage place Tesla and Skywell vehicles in the fleet as well, which means three different charging curves, three connector and software behaviours, and three maintenance regimes inside one operation. Standardising a fleet is a decades-long project; running a mixed fleet through a transition is the actual job. That is also why the 91% hybrid-or-electric figure deserves caution as well as credit. A fleet that is 91% electrified in the loose sense can still be a fleet where the daily miles are overwhelmingly driven by cars with combustion engines under the bonnet, because hybrids do the high-utilisation work and the battery-electric cars do the predictable work. The number that would settle it — electric-only share of fleet kilometres — is not one anyone publishes. 2040 is the target; 2030 is the test Two hundred and eight points, twelve to eighteen months, two hubs to start. Those are the numbers that will decide whether this works, and none of them is about 2040. The real deadline is 2030, when DTC expects to pass 2,500 electric vehicles. Two thousand five hundred cars sharing a network that is still being built is a much harder operational problem than 435 cars charging at two depots. It requires charging to be available where the taxis are, not where the company’s own property happens to be, and that means kerbside and rank-side infrastructure rather than depot infrastructure. The honest test is simple enough to run from outside. Watch whether the 208 points are actually energised within the stated 12 to 18 months. Watch whether the expansion figure of 354 by 2040 gets revised upward as the 2030 fleet target approaches, because if it does not, the arithmetic does not close. Watch whether any of the later sites are at ranks rather than depots. The driver in Muhaisnah 4 will tell you whether it worked long before any of that is published. If the cable goes in and he is back on the road inside half an hour, the programme is real. If he is waiting behind four other cars at shift change, it is a press release.

How the story usually gets told – SeaPRwire – The version that circulates goes like this. Malaysia decided that big social platforms had been operating in the country without permission for long enough, set a threshold of eight million local users, told anyone above it to apply for an Applications Service Provider Class licence — ASP(C), in the Communications and Multimedia Act 1998 framework — and switched the requirement on at the start of 2025. TikTok and WeChat complied and were licensed. X said it was below the line. YouTube argued about which category it belonged in. Order was restored. It is a tidy story and it contains a real event. The Malaysian Communications and Multimedia Commission announced the framework in mid-2024 with a 1 January 2025 start date, and the licences did start appearing at the beginning of that year. What the tidy version leaves out is that almost every number in it is unreviewable from the outside. Not wrong. Unreviewable. There is a difference, and the difference is the whole subject of this piece. Eight million, counted by whom? Start with the threshold itself. Eight million users in Malaysia. It sounds precise. It is the kind of figure that reads like a measurement and functions like a policy choice. Which is fine — thresholds are always policy choices — but it does raise the question that nobody in the public record has answered: counted how, by whom, and as of when? Monthly active users, or registered accounts? Accounts registered with a Malaysian phone number, or accounts observed accessing from Malaysian IP ranges? Deduplicated across devices, or counted per device? Measured by the platform, by the regulator, or by a third-party panel? Each of these choices moves a platform across the eight million line. A service with 6 million monthly actives and 11 million dormant registrations is above the threshold under one definition and safely below it under another. The distinction is not academic, because the licence obligations that follow are operational and expensive: age verification, local moderation capacity, response times on harm reports. If MCMC published its counting methodology, this would be a paragraph instead of a section. It has not, at least not in a form that lets anyone reproduce the arithmetic. What X told the regulator The X case is the cleanest illustration, because it is the one where a platform publicly disputed the count. X’s position, as reported at the start of 2025, was that it did not meet the threshold in Malaysia — citing a user figure of roughly 5.71 million from its own early-2024 reporting. MCMC responded that it was verifying the claim. The platform was not licensed at the point the requirement took effect. Pause on the structure of that exchange. The regulated party supplies the number. The regulator checks the number. The number determines whether the party is regulated at all. That is not a crazy arrangement — self-reporting with audit behind it is how a lot of financial regulation works — but it only produces trust when the audit produces a published result. So: has MCMC confirmed or rejected X’s figure? Not in any document I can find. Did X update the number? Not publicly. The file appears to have stopped at “we are verifying.” Meanwhile Google, for YouTube, took a different route entirely: rather than disputing the user count, it raised a classification question about whether video-sharing sits inside the licensing category as drafted. Telegram was reported to be at the final stage of the process. Meta was reported to have started applications for Facebook, Instagram and WhatsApp. Some of these platforms are almost certainly above eight million Malaysian users; none of them seems to have been measured into or out of the regime on the public record. Licence conditions still unposted Here is the gap that matters most for anyone actually trying to comply. An ASP(C) licence is not a badge. It comes with conditions, and those conditions are what a platform has to build against. Reporting on the framework describes the obligations in general terms: age verification with no access for under-13s, local content moderation teams able to respond to harm reports within 24 hours, proactive flagging of undisclosed sponsored content. Those are specific-sounding. They are also, as far as I can establish, paraphrases rather than published conditions. What is missing: The actual text of the licence conditions, per platform. Whether the conditions differ between a messaging service and a short-video service. The 24-hour response clock: measured from what event, against which categories of report, with what exemption. The audit cadence, and whether audit findings are ever published. The penalty schedule, and what triggers a suspension rather than a fine. This is not an accusation of secrecy for its own sake. Regulators routinely publish conditions and routinely withhold commercially sensitive detail. But a framework whose obligations are known mainly through secondary summaries produces a predictable outcome: platforms build to the summary, creators guess at the rest, and enforcement becomes unpredictable for everyone. A tragedy, then a timetable The licensing push did not arrive out of a policy seminar. The death of a young Malaysian TikTok creator, widely reported under the name Esha and attributed to sustained online harassment, produced an immediate government response in 2024 and is the event most often cited as the catalyst for both the ASP(C) framework and the strengthened penalties in the CMA 1998 amendment. It is worth being careful here, because tragedies get used to justify instruments that would not survive on their own merits. The question is not whether the response was sincere. The question is whether the instrument actually addresses the failure. Platform licensing does two things in principle. It gives the regulator a lever — a licence can be suspended — and it forces platforms to stand up local moderation capacity. Both are real. What licensing does not automatically do is produce a record. If a platform’s local moderation team handled 40,000 harm reports last year, how many were cyberbullying, how many took longer than 24 hours, how many resulted in account action, and how many appeals succeeded? That is the data that would tell you whether the lever is being pulled. It is also the data that no regulator in the region publishes as a matter of course. Where the enforcement record is thin Malaysia’s enforcement record is not empty. It is just pointed at individuals rather than platforms. The case usually cited involves a Malaysian influencer couple and roughly RM740,000 in funds frozen in mid-2024 in connection with an investigation into an alleged online donation scam. A separate line of enforcement has gone after undisclosed advertising: creators presenting paid endorsements as personal experience, particularly in the slimming and health product categories, where the Medicine Advertisements Board approval regime applies and where claims about treating or preventing disease are prohibited without substantiation. There is also a body of commercial disputes grounded in the same activities — a brand using an influencer’s image beyond the scope of a contract, for instance, which is a copyright and licensing question rather than a platform question. Notice what all of these have in common. They are downstream. They catch a person after the content has run. They do not tell you anything about whether the platform surfaced or suppressed the content, how fast it responded to a report, or whether a disclosure requirement was enforced at scale. Individual creators: visible enforcement, financial penalties, reputational consequences. Brands and agencies: contract disputes, occasional damages. Platforms: licensing status changes, and essentially no published service-level record. A regulator can argue that the first two are the deterrent and the third is the plumbing. A reader trying to assess whether the regime works cannot, because the third row is blank. There is a second asymmetry worth naming. Enforcement against individuals is cheap, fast and visible, which makes it attractive to a regulator under political pressure after a high-profile case. Enforcement against a platform is slow, technical and usually settled in a meeting rather than a courtroom. The first produces a press release. The second produces a compliance roadmap. A regime that optimises for the visible case will keep producing the visible case, and the platform-level behaviour it was supposedly built to change will keep drifting out of frame. That is not an argument against charging individuals who break the rules. It is an argument about what the enforcement mix tells you about what a regulator is actually optimising, and in Malaysia’s case the mix still looks like the one that existed before the platforms were licensed at all. Malay-language conversion, and the number behind it One statistic circulates constantly in Malaysian creator marketing: Malay-language content converts better than English. The figure most often attached to it is a 23% higher average engagement rate on Malay-language video, attributed to TikTok’s own data. It is probably directionally true. Malaysia’s population is majority Malay-speaking, code-switching between Malay and English is normal in urban areas, and a creator speaking the language of the majority audience will usually do better. It is also a platform-published number, and it is used by agencies to justify budgets. That should prompt the same question as the eight million threshold: what is the denominator, over what period, across how many accounts, and was it measured on engagement rate or on something closer to money? The commercially relevant figure is not engagement. It is order value per thousand views, and return rate, and the cost of a Malay-language content team against the lift. None of those appear in the publicly repeated statistic. A 23% engagement lift on a category with a 4% conversion rate is a different business from a 23% lift on a category converting at 0.4%. The same caution applies to the market-size numbers. TikTok was reported at more than 18 million Malaysian users in 2025 with daily usage above 75 minutes, both from third-party digital reporting rather than from the platform’s audited disclosures. Those are the best numbers available. They are also self-selected into the reporting that everyone cites, which is how a figure becomes true by repetition. SIRIM, LHDN, and the paperwork underneath The part of Malaysian content commerce that is genuinely well-documented is the boring part. Selling electronics into the country means SIRIM certification, with specific standards such as MS1865 for certain product categories. Skipping it does not produce a takedown notice; it produces a delisting and, potentially, a seizure. Taking money from brand work means declaring it. The Inland Revenue Board’s position is that influencer income is income, including in-kind: the PR product, the hotel stay, the voucher. Form B exists for a reason, and the creator who logs only cash is building a problem. Disclosure rules are more prescriptive than most creators assume. The working guidance across platforms in Malaysia: Instagram: the built-in Paid Partnership tag plus an #Ad or #Sponsored marker — the tag is not a substitute for the marker. TikTok: explicit commission disclosure for affiliate content, a stricter standard than for ordinary posts. YouTube: spoken and on-screen disclosure inside the video itself, ideally within the first two minutes; a description line alone does not satisfy the requirement. Health and slimming claims: substantiation, and in some categories Medicine Advertisements Board approval before any efficacy claim runs. Payments and logistics are equally local and equally unromantic. A Malaysian storefront that does not support GrabPay and Touch ‘n Go is leaving orders on the table; one that does not plan around Ramadan and Hari Raya is planning wrong. All of this is checkable, which is exactly why it is worth more than the headline numbers. It is the layer where being wrong costs money immediately. The 2025 amendment, read closely Alongside the licensing framework came amendments to the Communications and Multimedia Act 1998, described in coverage as strengthening penalties. That phrase does a lot of work and deserves picking apart, because “stronger penalties” can mean several different things and only one of them changes behaviour. Penalties can rise for the same conduct, which matters mainly if the old penalty was small enough to be written off as a cost of doing business. They can extend to new categories of conduct, which matters more. They can attach to new duty-holders, which matters most of all, because shifting liability from the individual poster to the platform is what actually changes what a platform builds. The licensing framework does the third thing structurally: a platform that can lose its licence for systemic failure will spend money differently from one that faces occasional fines. What is not clear from the public record is how the amendment interacts with the licence conditions — whether a breach of a condition is itself an offence, or whether the regulator’s route is suspension first and prosecution only for the underlying content. That sequencing is the whole ballgame for a platform’s legal team. Fine-then-suspend and suspend-then-fine produce completely different risk models, and they produce them before a single ringgit changes hands. Anyone who tells you the amendment “cracked down” on online harm should be asked which of the three it did. The answer, on the evidence available, is some of the first and an unclear amount of the third. Platform incentives changed, and not where creators expect The under-discussed consequence of licensing is that it realigns the platform against the creator. Before January 2025, an undisclosed sponsored post was a disagreement between a creator and a regulator, with the platform as venue. After licensing, the platform has skin in the game: keeping an ASP(C) licence requires demonstrable enforcement, and a visible pattern of undisclosed paid content is evidence that it is not enforcing. So the platforms now have a reason to surface the very thing that creators would rather stayed quiet. Reporting on the regime describes exactly this — proactive flagging of non-disclosed sponsored content as a licence obligation. A creator who treats disclosure as optional is no longer only betting against MCMC. They are betting against the platform’s own compliance team, which has a licence to protect and no particular loyalty to any individual account. Age verification: no access for under-13s, which changes how a creator can address a young audience and how a brand can target one. Local moderation with a response clock: the platform must be able to act on a harm report inside a day, which means more automated enforcement and less discretion. Proactive flagging of undisclosed ads: the platform’s incentive is to over-flag, not under-flag. Record-keeping: audit trails exist because the regulator may ask for them, and they cut both ways. That is a real change, and it is the part of the framework most likely to actually alter day-to-day practice. It is also, again, known through summaries rather than through published conditions. Three questions a brand should put in writing For the companies paying for Malaysian creator campaigns, the practical response is unglamorous: move the ambiguity into the contract. Most disputes in this area are not about whether a rule was broken. They are about who carried the risk of the rule being unclear. A brand that specifies the obligation and pays for the compliance step has a very different position from one that sends a product and a brief and assumes the creator handled it. Who files the disclosure, and where? Name the platform, the mechanism, and the wording. “Comply with local law” is not a specification. Who holds the substantiation file? For health, slimming and any claim adjacent to a medical condition, the evidence must exist before the content runs, not after a complaint. What happens to usage rights after the campaign? The image-reuse disputes in this market are contract disputes. Scope, territory, duration and paid amplification should each be a separate line. None of this requires knowing the licence conditions. It requires accepting that they are unspecified and pricing that in. What the takedown numbers would need to look like It is worth being concrete about what a real platform-level record would contain, because “transparency report” has become a phrase that means almost anything. A useful Malaysian platform record would be small. Five lines, published quarterly, per licensed platform: Reports received, split by category — harassment, fraud, child safety, intellectual property, everything else. Median and 90th-percentile time to first action on a harm report, with the clock defined. Items removed or restricted, as a count and as a share of reports received. Appeals filed, and appeals successful. Referrals to MCMC or to the police. That is the entire list. It is not commercially sensitive in any meaningful way; every large platform already computes most of these internally. The reason it does not get published in this form is that each line invites a comparison the platform would rather not have, and the regulator has not made it a licence condition. Without those five lines, every claim about whether the regime works is unfalsifiable. A platform can say it removed a great deal of harmful content, and nobody can check whether it removed content that was harmful or content that was merely inconvenient. A regulator can say enforcement has tightened, and nobody can check against what baseline. This is not specific to Malaysia. It is the standard condition of platform regulation almost everywhere: the instrument is built, the obligation is real, and the measurement is left to the regulated party’s goodwill. Malaysia’s distinction is that it built a licensing lever, which means it could attach publication duties to the licence — and, on the evidence available, has not done so in any visible way. The indicator to watch If you want to know whether this regime is working, do not watch the licence list. Watch for MCMC publishing a user-count methodology, even a thin one. Watch for the licence conditions appearing as text rather than as summary. Watch for the first published set of platform-level response statistics — reports received, resolved within 24 hours, appealed, overturned. Until at least one of those appears, the honest position is that Malaysia built a real instrument and has not yet shown anybody the readings. The licensing is not theatre. The missing scorecard is what keeps it from being evidence. And the eight million figure, which started all of this, remains what it has been since the day it was announced: a number that decides who is regulated, arrived at by a process nobody outside the commission can inspect.

A hydrant in Chengdu has a SIM card – SeaPRwire – Somewhere in Chengdu there is a fire hydrant that reports for duty every morning. It sends a short packet through a carrier’s network, several times a day, saying in effect that it is still a hydrant, still upright, still where the map says it is. The electricity meter in the same stairwell does something similar. So does the lift, the shared bicycle locked to the rack outside, and the parcel locker by the gate that has started charging tenants for the fourth hour of storage. None of these objects has a voice plan. None of them streams video. Each of them holds a SIM card and each of them counts, in the regulator’s arithmetic, as a connection. By the end of 2025 the three carriers in China (Mainland) — China Telecom, China Mobile and China Unicom — reported 2.888 billion cellular IoT terminal users, a net increase of 232 million over the year. That figure now exceeds the 1.061 billion mobile phone users on the same networks. Machines outnumber people on the Chinese mobile network by something close to three to one, and the ratio is still widening. IoT connections make up 61.3% of every terminal connected to those mobile networks. This is the part of the story that never makes the press conference. The other part does: 5G base stations, gigabit fibre, computing hubs in the interior. Both parts are true, and they are true in the same year, and only one of them moves revenue. The 9.1% and the 0.7% Here are the two numbers the Ministry of Industry and Information Technology put at the top of its 2025 telecommunications statistics bulletin, released in late January 2026. Telecommunications service volume — a quantity index, roughly a measure of how much traffic and how many connections the system carried, weighted and deflated — rose 9.1% year on year. Telecommunications service revenue rose 0.7%. Read those two lines next to each other for a while. Nine point one against zero point seven. A network that carried almost a tenth more of everything it carries, and brought in well under one percent more money for doing it. (The carriers will point out, reasonably, that volume and revenue measure different things, and that a deflated volume index is not a price. It is still a fairly stark pair of numbers to publish about yourself.) This gap is not a Chinese peculiarity. Every operator in every market that built out fibre and 5G in the 2020s is living inside some version of it. What is distinctive about China (Mainland) is the scale at which the gap is now running, and the frankness with which the regulator puts both figures in the same bulletin. The practical consequence lands on the capital side. If traffic grows at high single digits and revenue grows at a rounding error, then the return on the next base station is not the return on the last one. The engineering answer has been to chase cost out of the network. The commercial answer, so far, has been to sell connectivity in units nobody would have recognised as a product in 2015: a hydrant check-in, a lift diagnostic, thirty seconds of video from a dashcam. Where the base stations sit The build-out did not slow down. At the end of 2025 the country had 12.87 million mobile base stations in total, up 227,000 on the year. The split inside that number is more interesting than the total. 5G base stations: 4.838 million, a net addition of 588,000, now 6% of all mobile base stations, up four percentage points on the year. 4G base stations: 7.192 million, up only 80,000. 4G is still the larger single technology by site count, and it will be for a while. Base stations with 5G RedCap capability: 2.064 million, equal to 7% of all 5G sites. That last line is the one that explains the hydrant. RedCap — Reduced Capability, the industry’s own name for the cut-down version of 5G — is designed for devices that need more than a legacy IoT chip and far less than a phone. Lower bandwidth, lower power, cheaper modules, no need for the full antenna count. Rolling it out across 42.7% of the 5G estate within roughly a year of it becoming a deployment priority is the carrier’s direct answer to the margin question: if the network is going to carry 2.888 billion things, then a large share of those things should occupy the cheapest possible slice of it. On the user side, 1.204 billion 5G mobile subscribers were counted at the end of 2025, 65.9% of all mobile phone users, up 9.2 percentage points in twelve months. The migration is far enough along that the remaining 34% is now the harder half: older devices, price-sensitive users, rural households whose 4G handset still does everything they ask of it. RedCap, the cheaper 5G It is worth slowing down on RedCap, because it is the clearest example of how the Chinese network shifted from a consumer build to an industrial one. A full 5G module in a sensor is absurd. You do not need multi-gigabit downlink to tell a district heating utility that a valve is closed. What you do need is a module cheap enough to deploy in tens of millions, a battery that lasts years rather than days, and a network that will still be maintained in 2035. RedCap trades bandwidth and latency headroom for exactly those three things. The numbers on the ground are straightforward. Out of 4.838 million 5G base stations, 2.064 million had been upgraded to accept RedCap devices by the end of 2025. That is not a pilot. That is a coverage decision made at national scale, and it is what makes the 28.88 million-per-year growth in IoT connections physically possible rather than aspirational. Meters, hydrants, lifts and shared mobility: cheap modules, tiny payloads, years of battery life. Dashcams and in-vehicle units: intermittent bursts, moderate bandwidth, high device counts. Wearables and health monitors: low duty cycle, long replacement cycles, tight power budgets. Factory sensors and inspection cameras: the segment that actually justifies 5G’s latency claims. The fourth category is the one the policy documents care about, and the one that is hardest to count. The first three are the ones that show up in the connection statistics, and they show up in enormous numbers. Fibre, 10G PON, and 74.99 million kilometres of glass Fixed infrastructure kept the same rhythm. During 2025, 2.113 million kilometres of new optical cable were laid, taking the national total to 74.99 million kilometres. For scale, the equator is about 40,000 kilometres, so the country now has enough glass in the ground to circle it roughly 1,875 times. (This is the kind of fact that appears in every national telecom bulletin and means very little, which is precisely why it appears in every national telecom bulletin.) The useful numbers sit under it: Fixed broadband access ports: 1.251 billion, up 48.77 million. Fibre access (FTTH/O) ports: 1.21 billion, up 50.30 million, now 8% of all broadband ports. 10G PON ports capable of gigabit service: 31.62 million, up 3.419 million. Fixed broadband subscribers: 691 million, up 20.99 million. Subscribers on 1000 Mbps or faster: 238 million, up 31.57 million, 5% of all fixed broadband users. Copper is functionally finished as an access technology. At 96.8% fibre ports, the remaining 3.2% is legacy estate being retired rather than a competing platform. The gigabit number deserves a second look. Thirty-one point six million 10G PON ports serve 238 million gigabit subscribers, which is a ratio of roughly 7.5 subscribers per port — a reasonable split ratio for passive optical networking, and one that suggests the port build is running slightly ahead of subscriber demand rather than desperately behind it. Growth of 31.57 million gigabit subscribers in a year, against 20.99 million total broadband subscriber growth, means the gigabit tier is now where essentially all fixed broadband growth happens. People are not joining; they are upgrading. Eight hubs, five of them in the west The computing side is where national policy became physical concrete. The “East Data, West Computing” programme, known in Chinese as dongshu xisuan, was designed to move data processing away from the coastal provinces that generate it and toward the interior provinces that have cheap land and, more importantly, cheap electricity. Eight national computing hub nodes were designated. Five of them are in the western region. The logic is not subtle: a training cluster or a cold-storage archive does not need to sit next to its users, it needs power that costs less and does not brown out in August, and land that a coastal municipality would rather use for something with a higher tax yield. By mid-2025, the National Data Administration was reporting total computing capacity as second in the world by scale, alongside a claim about the geographic distribution that matters more than the ranking: interior nodes were taking a growing share of new capacity, and the coastal provinces were, for the first time, not the default answer to “where does this data centre go”. Across the three carriers, 938,000 data centre racks were in service and offered externally at the end of 2025, up 108,000 on the year. That is a growth rate of about 13% in rack count, in a year when service revenue grew 0.7%. The description in the ministry’s own bulletin is revealing: the carriers’ focus has shifted from “broad coverage” to “deep integration”, from selling basic cloud resources to selling scheduling, and from building racks to making the racks they already have work together. That shift is a margin story. Racks are capital. Scheduling is software. If you can sell the same rack twice by filling its idle hours, you do not need to build as many racks. Racks, power, and the bill that arrives later There is a cost to all of this that the connection statistics do not capture. A 938,000-rack estate draws a great deal of electricity, and in China (Mainland) it draws it from a grid that is simultaneously absorbing an enormous build-out of solar and wind. The interior hubs were sited for cheap power, but “cheap power” in a province that is also industrialising fast is a moving target. Several western hub sites have already had to plan substation capacity alongside rack capacity, which is the sort of detail that turns a two-year project into a four-year one. The second cost is the one that arrives later: depreciation. A base station built in 2020 is on a different depreciation clock from a RedCap upgrade installed on it in 2025, and a rack commissioned in a hub in 2023 is competing with racks commissioned in 2025 that are denser, better cooled and cheaper to run. The carriers’ capital budgets have to absorb both, and the revenue line grows at 0.7% while they do it. Network traffic and connections: up sharply, and the volume index says 9.1%. Revenue per unit of traffic: down, and the revenue line says 0.7%. Capital intensity: still high, because coverage commitments do not expire. Power and land: the input cost that the western hubs were supposed to solve, and partly have. Put those four together and you get the actual strategic problem of Chinese telecom in the late 2020s. It is not a coverage problem and it is not a technology problem. It is the problem of an industry that has built something extremely good and has not yet found a way to charge for it in proportion to how good it is. The west connected faster than the east One regional detail in the bulletin deserves more attention than it got. Gigabit subscriber penetration at the end of 2025, by region: Western region: 36.1% of fixed broadband users, up 4.4 percentage points. Eastern region: 34.6%, up 3.4 points. Central region: 33.9%, up 2.7 points. North-east region: 29.0%, up 5.4 points. The west leads the east on gigabit penetration, and gained faster. There is a mundane explanation: the western provinces were starting from a lower base, and new build is cheaper to do at gigabit than upgrading an estate that already has copper-era plant in it. There is also a policy explanation, which is that the same programme that put five of eight computing hubs in the west has been pushing fibre and transmission spending there for several years. Revenue, though, has not followed capacity eastward. Telecom service revenue share by region in 2025: eastern 50.9%, western 24.3%, central 19.7%, north-east 5.1%. The Yangtze River Delta on its own accounts for 23.2% of national telecom revenue; the Beijing-Tianjin-Hebei region, 9.1%. So the network that is being built fastest is, in relative terms, being built where the money is thinner. That is a defensible national policy and a difficult commercial one, and the tension between the two is not going to resolve on its own. Who pays for a network that keeps getting cheaper Follow the money and the answer is: industry, mostly, and increasingly industry that is not telecom. The 5G applications that show up in the ministry’s framing are industrial — connected factories, port automation, mining, grid inspection, and the “5G + industrial internet” category that has been in every five-year document since 2019. These are the applications that justify the latency and the slicing, and they are also the applications that require someone other than a carrier to do the integration work. That is where the revenue shows up, if it shows up at all: not in the access charge, but in the system integration contract, the platform fee, the maintenance agreement. The carriers have spent a decade becoming very good at building networks and are now trying to become credible at selling everything wrapped around them, which is a different business with different margins and different competitors. For the consumer side, the picture is simpler and less comfortable. Prices have fallen, speeds have risen, and 65.9% of mobile users are on 5G. Almost none of those users is paying a premium for it. The upgrade was absorbed into existing plans, which is exactly what the 9.1%-versus-0.7% pair describes from the other end. 1.19 billion HarmonyOS devices and 60% of AI patents Two more figures from the same August 2025 State Council briefing are worth holding carefully, because both are large and both need a definition attached. The first: the HarmonyOS ecosystem passed 1.19 billion devices, spanning more than 1,200 product categories — phones, cars, appliances, and a long tail of things that used to ship with a proprietary little operating system nobody updated. That is a genuine installed base, and it is the largest single piece of evidence that the country has an alternative client platform with real scale. The second: 60% of the world’s artificial intelligence patents. This one is more slippery. Patent counts depend entirely on which office you count, whether you count applications or grants, and whether a family counts once or once per jurisdiction. Sixty percent of global AI patent filings is a defensible statement about filing behaviour at the national office. It is not a statement about who has the best models, and the briefing did not claim it was. The reason both numbers belong in an article about telecom infrastructure is the same reason RedCap belongs in it. The build-out was never only about access. The stated goal, repeated across four years of policy documents, has been a domestic stack: domestic chips, domestic operating systems, domestic models, running on domestic networks, sold to domestic industry. Access layer: 4.838 million 5G sites, 1.21 billion fibre ports, 31.62 million 10G PON ports. Device layer: 1.19 billion HarmonyOS endpoints, 1.204 billion 5G subscribers. Compute layer: eight national hubs, five in the west, 938,000 external racks. Application layer: 2.888 billion IoT connections, 61.3% of everything attached to the mobile network. Each layer is measured by a different agency, in different units, on a different schedule. That is precisely why the revenue figure is the one worth watching — it is the only number in the stack that cannot be flattered by choosing a unit. What the bulletin declines to break out A national statistics bulletin is a political document as well as a technical one, and its silences are informative. There is no breakdown of IoT revenue per connection. There is no average revenue per user for the gigabit tier, only subscriber counts and port counts. There is no utilisation rate for the 938,000 racks, which is the single number that would tell you whether the western hubs are filling or standing half empty. There is no figure for how much of the 588,000 net new 5G sites were built to meet a coverage obligation rather than a demand signal. None of these omissions is sinister. All of them are commercially sensitive, and a regulator that published them would hear about it. But a reader trying to work out whether the 9.1% versus 0.7% gap is narrowing or widening cannot do it from this bulletin alone, because the two lines that would settle it — revenue per IoT connection, and rack utilisation by hub — are exactly the two that are missing. That leaves outside observers doing arithmetic on the numbers that are published. Divide 232 million net new IoT connections into the difference between total service revenue growth and what subscriber growth alone would explain, and you get a number with an enormous error bar and the right sign. It suggests IoT is being sold at a fraction of the per-unit revenue of a phone line, which is unsurprising, and that the fraction is not improving quickly, which matters. Read the revenue line, not the press release If you want to know how this goes, do not read the base station count. Base stations are a commitment, and commitments get met. Racks are a commitment too. What is not a commitment is the revenue line, and that is where the honest information lives. Watch three things next year: whether service revenue growth climbs back above two percent, whether IoT connection growth continues to outrun any plausible growth in IoT revenue per connection, and whether the western hubs start reporting utilisation figures rather than capacity figures. The hydrant in Chengdu will keep checking in either way. It has no opinion about any of this, which is of course the entire point of it.

AsiaPresswire – The numbers tell one story and the streets tell another. In the first full year after the borders reopened, Thailand welcomed visitors in the tens of millions again, a figure that travels well in headlines and government press releases. Walk the strip near a Bangkok transit hub and the composition of that number is visible in a way no statistic captures. The tour groups are smaller. The languages at the food stalls are different. The traveler with a laptop and no return ticket is more common than the traveler with a camera and a seven-day itinerary. Thailand’s tourism rebound is real. The tourist it was built for is not the one who showed up. Spend a morning at any transit hub and the shift is visible before you read a single report. The family of four with matching luggage and a tour flag is still there, but they share the pavement with a young woman in headphones answering email between train and co-working space, and with a retired couple who know the neighborhood better than the concierge does. The country built its entire service layer — the buses, the banquet halls, the group menus — around the first type. The second and third arrived anyway, and the layer is still catching up. The timeline that everyone remembers wrong The pandemic shut the country’s borders for the better part of two years, and the story most people carry is a simple one: closed, then open, then back to normal. The real sequence was messier. There was a slow trickle of sandbox programs and quarantine hotels, then a hesitant reopening weighted with testing, then a sudden drop of almost all requirements that released a dam. Each step changed who could come and who bothered to. The visitors who returned first were not the mass-market charter crowds of 2019. They were the attached ones — expatriates, second-home holders, and long-stay regulars who treated the reopening as a homecoming. The leisure fly-in masses followed later and in a different shape. By the time the headline arrivals number approached the old peak, the mix underneath it had already been rewritten. The first wave back were residents and regulars, not tourists. The middle wave came from nearer markets that could fly short-haul. The late wave looked like the old one only in total count, not in composition. What a hotelier in Chiang Mai noticed A hotel owner I spoke with runs a mid-size property in Chiang Mai, far from the Phuket beach crowds. She described the change in concrete terms. Before 2020, her winter bookings were dominated by a few large Chinese tour operators who filled buses and banquet halls. After reopening, those operators came back slowly and in smaller blocks. The rooms they left were taken by a different guest: a remote worker staying six weeks, a retired couple on a three-month lease, a family that treated the city as a base rather than a destination. “The bus groups paid less per head but they were predictable,” she said. “The long-stayers pay more and stay longer, but I can’t fill a banquet hall with them, and I never know which month they’ll pick.” Her revenue recovered faster than her occupancy math would suggest, because the new guest spends differently — on laundry and rentals and cooking classes, not on group dinners and bus excursions. The hotel did not go back to normal. It went somewhere new. The street-food vendor’s view The change shows up most honestly at the stall level. A vendor near a night market told me his regulars used to be a rotating cast of package tourists who ordered one pad thai and took a photo. Now his steady trade comes from people who live nearby for a season — digital nomads, language students, and the new class of long-stay visa holders who cook less and eat out more. He adjusted without a memo. He added a smaller portion size for solo diners, learned to take a payment app the nomads use, and started opening an hour later to catch the late remote-worker dinner crowd. His income held, but the rhythm changed. “The old tourists came in waves,” he said. “These ones come every day, then vanish for two months and come back.” For a small vendor, a daily regular who disappears is harder to plan around than a tour bus that always arrives at six. The shift away from one market For two decades Thailand’s tourism was built, quietly and heavily, on one source market. That market’s travelers filled hotels, malls, and tour coaches across the country. When it contracted during the border years, the dependence became a liability, and the recovery has deliberately diversified. The new arrival mix leans more on nearer neighbors, on a fast-growing market to the southwest, and on a scattered set of long-haul guests who stay for months at a time. This diversification is the structural story beneath the rebound. A country that once rode one wave is now learning to ride several smaller ones. The total can look similar on a spreadsheet, but the resilience is different, because no single market can starve the whole system the way the old one could. Nearer neighbors fill the short-haul gap the old market left. A fast-growing southwestern market adds volume the old mix lacked. Long-haul long-stayers add stability that group tourism never did. The visa that changed who stays One quiet lever deserves more credit than it gets. The country introduced a visa route aimed at exactly the long-stay, laptop-carrying traveler, and it redrew the map of who settles in for a season. Where a tourist once arrived for two weeks and left, a remote worker can now arrive for months, living in a condo, renting a scooter, and becoming a regular at the same three cafes. That guest does not show up in the “tourist” imagination, but she shows up in the rental market, the language-school enrollment, and the neighborhood clinic. The effect compounds. A few thousand long-stayers pull in services — co-working spaces, visa agents, fitness studios — that then attract more of the same. A district can shift from package-tour stop to live-work base without any single decision declaring it. The visa was the match, not the fire, but it lit a corner that keeps burning. A guide who rebuilt his business A freelance guide in the old Chinese-tour circuit lost almost everything when the borders closed. He did not wait for the groups to return. He learned the language of a new market to the southwest, built a small network of apartment landlords, and started offering a different product: not temple buses, but week-long stays that mixed light guiding with help settling in. His income is steadier now than in the peak year, because his clients come for a month and need him the whole time, not for a single afternoon. “Solo travelers ask better questions than bus groups ever did,” he said. “They want the noodle shop the locals use, not the one in the brochure. I had to become a local again, not a narrator.” His pivot is the rebound in miniature: the same country, a different guest, and a worker who had to reinvent the job to meet them. The past, the present, and the next version Look at three slices of the same industry. In the past, Thailand sold a cheap, sun-and-sand package built for volume. In the present, it is selling a longer, slower stay built for value per week, not bodies per bus. In the next version, the edge may be something harder to photograph: a place that lets a visitor actually live somewhere for a season without pretending to immigrate. The danger in the next version is complacency. The old model was easy to count and easy to subsidize with group discounts. The new model is diffuse, harder to measure, and easy to neglect while the arrival number looks healthy. A government that watches only the total will miss the shift until a market moves and the base proves thinner than the headline suggested. Why the old playbook lingers Despite the shift, much of the industry still markets the old product, because the old product was easy to sell and easy to finance. Group packages have set margins and familiar buyers. Long-stay offerings are bespoke, slower to fill, and harder to pitch to a bank. So a hotel may quietly host month-long guests while still advertising week-long deals built for a visitor who no longer arrives in the old shape. The mismatch between what is sold and what is booked is one of the quieter stresses in the rebound, and it resolves only when the balance sheet catches up to the street. The part that does not photograph This is the piece of the rebound that a press release cannot show. A government can publish an arrival line crossing a peak. It cannot easily show that the line is now held up by several thinner ropes instead of one thick one, and that those ropes are made of visa terms, local rents, and air quality that no tourism ministry fully controls. The resilience is real. So is the fragility it replaced, only distributed and quieter than the old single-market cliff. A conversation about the season I sat with the Chiang Mai hotelier again as the high season approached. She was optimistic and wary at once. Her: “The bookings look like 2019 on the screen. But open the file and it’s a different guest list.” Me: “Does that worry you?” Her: “It worries me less than 2021 did. Back then I had one market and it was gone. Now I have six markets and they arrive in different months. If one stays home, the others fill the bed.” That, more than any arrival statistic, is the real story of the rebound. Not the height of the wave, but the number of waves, and the fact that no single one can drown the shore. The second-home crowd Many of the visitors who never left are not tourists at all in the traditional sense. They are expatriates, retirees, and remote workers who bought or leased a base in the country years ago and simply reabsorbed the travel stop into a life. The border closure forced a pause; reopening let them return to a home that happened to sit in a holiday destination. Their spending looks like tourism in the statistics and like residency in the neighborhoods. They are the ballast of the rebound — present in every season, indifferent to the headlines, and invisible to the bus-tour operator who still waits for the group that may never return in the old shape. The budget carrier’s quiet role Underneath the headline arrivals sits a quieter engine: the low-cost carriers that knit the nearer markets to Thailand’s airports. When the legacy long-haul charter model weakened, the short-haul budget flight filled the gap, bringing guests who could decide on a long weekend and change their minds by the month. This is not glamorous infrastructure, but it is the reason the recovery did not depend on a single distant market recovering first. The cheap seat turned proximity into volume, and volume into the diversified base the country now relies on. A carrier adding a route from a nearby capital does more for the resilience of Thai tourism than any billboard in a far-off city. What the future visitor wants If the past visitor wanted a checklist — a temple, a beach, a photo — the present one wants a place to be. The difference sounds soft, but it reshapes what gets built. The future guest is less interested in the buffet and more interested in the noodle shop, less in the guided bus and more in the rented scooter, less in the resort enclosure and more in the ordinary street. Thailand’s advantage is that it has the ordinary street in abundance. The countries that will lose the next version of this visitor are the ones that package the ordinary until it is no longer ordinary. The risk in the new model The diversified, longer-stay model is more resilient, but it is not risk-free. It depends on visa rules staying friendly, on the cost of living staying low enough to attract the long-stayer, and on the neighborhoods not pricing out the very charm that draws them. Each of those can move without warning. A visa tightening, a rent spike, or a stretch of bad air quality can send a cohort of long-stayers elsewhere in a season. The old model failed dramatically when one market vanished. The new model can fail quietly, one lease at a time, and quiet failure is harder to see until the beds stay empty. The beach and the city diverged Thailand is not one tourism market. The island resorts and the northern cities recovered on different clocks and with different guests. The beach destinations, built for the sun-and-sand package, snapped back faster in raw arrivals because the product was simple and the flights short. The northern cities, more dependent on the long-stay and overland traveler, recovered slower but changed more. A hotelier on the coast told me her property looked like 2019 by the second season, but her guest mix had tilted toward nearer neighbors who come for a week and come often, rather than the long-haul visitor who came once and stayed a month. The north, by contrast, gained the month-long guest and lost some of the week-long one. The country’s rebound is the sum of these two different recoveries, and pretending they are one number hides the shift happening inside it. The money that actually lands The economic question is not how many arrive, but how much of their spend stays local. A group tourist on a packaged deal spends a large share before arrival — the flight, the hotel block, the tour — and a smaller share on the ground, often at businesses linked to the operator’s home office. A long-stayer spends a greater share locally: rent, groceries, cafes, clinics, weekend trips. The same arrival can mean very different things for a provincial economy. This is why a smaller number of long-stayers can rival a larger number of fly-in guests in local impact. The guest who stays six weeks pays six weeks of rent to a local landlord and eats six weeks of dinners at local stalls. The guest who stays six days pays one. The headline counts them equally. The neighborhood does not. The comparison the government avoids Officials naturally compare today’s arrivals to the old peak and declare victory when the lines cross. That comparison is comforting and incomplete. The old peak was built on a single large market and a short-stay model. Today’s number rests on several smaller markets and a longer-stay model. Crossing the line in total is not the same as rebuilding what was lost, because the lost model was fragile and the new one is still being assembled. A government that reports the crossing without the composition tells half the story and the easier half. The arrivals number hides the undercount Official arrivals miss a layer of movement the statistic cannot see. Same-day visitors crossing by land, travelers who step out and step back for visa runs, and the small share who overstay and fade into the long-stay underground all live outside the clean total. None of this is unique to Thailand, but it matters here because the long-stay story is precisely the part the count underrepresents. The country may have more resident-like visitors than the arrival line suggests, which cuts both ways: more stable demand, but also more pressure on services built for transients. A district that flipped In one Chiang Mai neighborhood, the change is visible block by block. A row of guesthouses that once hosted package tourists now leases monthly to remote workers. A travel-agency storefront became a visa and co-working desk. A tour-bus parking lot became scooter rentals. No single policy flipped the street. The guests did, one lease at a time, and the businesses followed because surviving meant serving who actually showed up. The district did not rebrand. It was reoccupied. The strain nobody mentions A longer-stay model is not purely good news for the people who live there year-round. More foreign residents push up rents in the neighborhoods they favor, and the same charm that draws the long-stayer — quiet lanes, cheap eats, slow pace — erodes as landlords convert housing to short and medium lets. A local I spoke with, born and raised in the city, now commutes farther because the apartment she grew up near rents to a six-week visitor instead of a local family. The rebound helps the balance sheet. It complicates the block. Long-stayers raise local rents in the areas they cluster. Housing converted to medium lets squeezes year-round residents. The charm that attracts them is the first thing their arrival erodes. What a smart operator does now For a business deciding how to serve the new Thailand, the old playbook of volume discounts and group packages is necessary but not sufficient. The operators who thrive read the shift and adjust. Build for the week and the month, not just the night; offer weekly rates and settle-in help. Learn the languages of the growing markets, not only the legacy one. Diversify the source list so no single market can empty the beds. Treat the long-stay guest as a local customer, because that is what she is. The lesson underneath all of it: the tourist Thailand trained for is not the one who came back. The country that serves the one who stayed will outlast the one that waits for the one who left. The tourists who never left The headline recovery number will keep climbing, and the government will keep celebrating it, because arrivals are the metric everyone understands. But the traveler who matters most to Thailand now is the one who arrived, liked it, and simply did not go home on schedule. She is a teacher in a condo, a coder in a cafe, a retiree who renews a lease instead of a flight. The rebound is real. The tourist is not who the country trained for, and the country is quietly better off for it. The visitors who changed are the ones who stayed.

AsiaPresswire – A single number explains why the world keeps mistaking Saudi Arabia’s construction market for a certainty. In one recent year the kingdom awarded construction and infrastructure contracts worth well over a hundred billion dollars, a sum larger than the entire annual building output of most countries in the region. That figure, repeated in every briefing and pitch deck, creates a comforting illusion: that the cranes are the outcome, and the outcome is settled. The cranes are real. The certainty is not. The building boom is genuine, and it is the largest coordinated construction program on the planet this decade. But the gap between what gets announced, what gets contracted, and what actually opens is wide, and that gap is where the real story lives. Anyone planning around the headline number without reading the fine print of delivery will be surprised, and not in a pleasant way. The projects that dominate the imagination A handful of developments sit at the center of the story, and each one is sized to intimidate. NEOM, the planned zone in the northwest, carries a headline price tag in the hundreds of billions and a list of ideas — a linear city with no cars, vertical farming, artificial snow — that read like science fiction written by a planning department. The Red Sea Project promises a string of luxury resorts on previously empty coastline. Qiddiya aims to become a destination for entertainment and sport outside Riyadh. The capital itself is ringed by metro lines, stadiums, and a massive downtown expansion. These are not renderings in a brochure. Ground has moved, contracts have been let, and thousands of workers are on site. But scale and certainty are different things. A project announced at half a trillion dollars is not half a trillion dollars of finished building. It is a target, a negotiating position, and a statement of intent, all at once. The number tells you the ambition. It does not tell you the delivery date. A useful detail: most of these developments are being built in phases, and phase one is almost always smaller than the full vision. A resort might open with a handful of hotels rather than the planned dozens. A new district might deliver its first cluster of towers while the rest remains graded earth. Phasing is not a retreat from the vision. It is how the kingdom manages the gap between announcement and demand, opening what it can fill and holding the rest until the habit forms. The investor who reads phase one as the whole project misreads both the scale and the timing, and bids as if the renderings were a schedule rather than a wish. A project manager who bet on the boom To understand the pace, talk to someone living inside it. A civil engineer I know left a comfortable job in Dubai and moved his family to Riyadh because, as he put it, “the work here is once in a generation, and Dubai was winding down its big cycle.” He arrived expecting a straight line from contract to completion. He found a different shape. His first assignment was a mid-size commercial tower tied to one of the capital’s expansion programs. The design was approved quickly. The permit came slower than he expected. The steel arrived later than the schedule promised because a regional supplier had overcommitted across three giga-projects at once. The finishes slipped by months. None of it was catastrophic. All of it was normal. “Everyone thinks these projects are late because someone failed,” he told me. “They’re not late. They’re on the only schedule the physics and the supply chain allow. The brochure just promised a faster one.” That sentence is the key to the whole market. The announcements run on ambition. The sites run on physics. What the local-content rules actually do One force shaping the boom that outsiders miss is the push for local content. The kingdom ties a share of project spend to goods and services sourced inside the country, and it scores bidders on how much they will hire locally and buy locally. On paper this is industrial policy aimed at building a domestic supply base that outlasts the construction cycle. In practice it changes how projects are built. A contractor who could source steel or elevators cheaply from overseas now has a reason to look inland first. That builds local capability, but it also means the supply chain is being built at the same time the projects are being built, which is slower than plugging into an existing one. The rule is rational. It is also a tax on speed, paid in the early years of the program. Local-content scoring rewards bidders who hire and buy inside the kingdom. The domestic supply chain is being built alongside the projects, not before them. The speed tax is real in the early phase and eases as local capacity matures. The labor math nobody celebrates Behind every tower is a workforce equation that rarely makes the press release. The giga-projects need huge numbers of skilled and unskilled workers, and the kingdom imports most of them. Housing them, moving them, and keeping them on site through extreme summer heat is a logistics operation as large as some of the buildings themselves. The math gets tight in two places. First, the projects compete for the same scarce trades — electricians, crane operators, finishing crews — and when three megaprojects hire from the same shallow pool, wages rise and schedules slip for all of them. Second, the push to employ more Saudi nationals in skilled roles is real and necessary, but it takes time to grow those workers, and the gap is filled by expatriates in the meantime. The boom is therefore labor-constrained at the top, not just at the bottom. A useful way to see it: the bottleneck is not concrete. It is the people who pour it, finish it, and inspect it. The country can buy the material. It cannot instantly buy the hands. Where the money actually comes from The scale of the building program raises an obvious question: who pays? The answer is a stack, not a single source. The sovereign wealth fund backs the giga-projects directly. The state budget funds infrastructure and utilities. Local and regional banks extend project finance. International firms join as partners and contractors, often carrying part of the risk. That stack is sturdy, which is why the boom is not a bubble in the simple sense. But it is also concentrated. When the same fund and the same treasury stand behind most of the marquee work, a shift in priorities at the top moves every project at once. The risk is not default. The risk is reprioritization — a flagship slipping because a newer flagship took its place in the queue. The resorts and the reality of empty coastline The Red Sea and similar coastal developments tell a quieter story about demand. Building the resort is one task. Filling it with paying guests, year after year, is another. The kingdom is betting that a new class of traveler will choose its coastline over established destinations in the Mediterranean and the Indian Ocean. The bet may well pay, but the timeline is longer than the construction schedule. A manager involved in one coastal launch described the gap plainly. “We can open a resort in thirty months. We cannot manufacture a travel habit in thirty months. That habit is built by airlines, by visa rules, by a decade of word of mouth.” The buildings finish on the project clock. The demand finishes on a slower one, and the two clocks rarely sync. Which developments will actually open Not every announced development becomes a finished place, and the market knows it. The ones most likely to open on something close to plan share a few traits. They serve a function the kingdom needs regardless of tourism fashion — a metro line commuters use, a stadium for a tournament already awarded, a government district that consolidates workers already employed. They are anchored by use, not just by vision. The ones most likely to stall are the pure statements: the farthest-flung concepts, the most experimental forms, the ideas that depend on a future traveler or resident who has not been convinced yet. Those are not doomed. They are simply the ones whose completion date is set by conviction rather than by need, and conviction is the weaker scheduler. Projects tied to existing use finish closer to plan than projects tied to hoped-for demand. The metro and the stadiums have hard deadlines from events already committed. The speculative coastline depends on habits that take a decade, not a season. What a realistic operator plans around For a firm deciding whether to chase this market, the lesson is to plan around delivery, not announcements. Bid on the work whose completion is backed by a function, not just a flag. Build the local relationships the content rules reward. Hire for the labor constraint, not against it. And price the speed tax into every schedule, because the early-phase program runs slower than the brochure implies. The boom will continue. The kingdom has the will, the money, and the workforce pipeline to keep cranes in the air for years. But the skyline that photographs so cleanly hides a messier truth about pace. The buildings are coming. The question worth answering is not whether, but which ones, and on whose clock. The size in plain terms To feel the scale, put the headline number next to something familiar. A single recent year of awarded construction contracts in the kingdom ran into the hundreds of billions, larger than the entire annual building output of most of its neighbors combined. One planned zone alone carries a public price tag that, if spent in full, would rival the gross product of a mid-sized country. These comparisons are not precise, and the figures are targets more than commitments, but the order of magnitude is what matters. This is not a market with one big project. It is a market where the big projects are the baseline. The deadlines that force some projects to finish Unlike a resort that can open whenever demand allows, a metro line tied to a committed event has a date that does not move. When a global sporting event is awarded to a city, the transport that serves it becomes non-negotiable, and the schedule bends the rest of the program around it. The same logic holds for stadiums and the districts built to house the people who run those events. These are the projects most likely to open on time, because missing the date is visible to the entire world, not just to a balance sheet. This is why a seasoned operator separates event-anchored work from vision-anchored work. The first carries an external clock it cannot ignore. The second carries only the internal clock of ambition, and that is the clock that slips. Event-anchored projects have a date set outside the program, so they resist slippage. Vision-anchored projects slip quietly, because no outside force enforces their date. The operator who tells them apart bids the first and watches the second. Foreign contractors and the risk they price in Almost none of the marquee work is built by a single local firm. It is built by international contractors, often in joint ventures with local partners, and the risk allocation is where the real negotiation happens. A foreign firm bidding on a giga-project prices in currency exposure, material inflation, and the cost of a schedule it does not fully control. Those costs land in the contract, and they are one reason the headline project value and the eventual spend diverge. The joint-venture structure exists for a reason. The foreign partner brings delivery method and skilled supervision. The local partner brings access, relationships, and a way to satisfy the content rules. Together they move faster than either could alone, but two management cultures on one site is its own small tax on pace. The housing that has to exist before the offices fill A downtown expansion only works if people live near it. The kingdom is building offices and entertainment districts at a pace that assumes a growing population will occupy them. The population is growing, but not always where the buildings go, and not always at the speed the buildings arrive. A new commercial district can stand finished and half-empty because the housing and transit that feed it opened later, or because the workers who were meant to fill it commute from far away. This is the quiet risk in the boom: supply of space arriving ahead of the demand that justifies it. The bold projects get announced together, but they complete on separate clocks, and the mismatch leaves gaps that take years to close. Operators who plan around the gap, rather than the rendering, avoid betting on a skyline that fills on hope. The supply chain built twice A subtle cost few briefing decks show is that the local supply chain is being constructed at the same moment the buildings are. Concrete plants, steel fabrication, elevator servicing, and finishing trades are scaling up while the projects they serve are also scaling up. In an established market those suppliers already exist and simply take the order. Here they must be created, trained, and equipped first. The result is a program building its own ladder while climbing it. That dual build is why early-phase costs run higher and schedules run longer than later phases will. The second half of the boom will look smoother than the first, not because the ambition shrank, but because the ladder finally exists. Summer and the clock The climate imposes a clock of its own. The summer months push temperatures to levels where outdoor heavy work must shift to nights or slow down, and that compresses the productive window for site labor. A project that plans its concrete pours and steel lifts around the cool season moves faster than one that assumes a uniform year. The engineers who thrive here treat the calendar as a material constraint, not a formality, and they schedule the critical path to dodge the worst of the heat rather than fight it. What a stall looks like up close Not every site that slows down stops. A stall often looks like a half-finished structure that keeps its crew but loses its momentum, a phase of work waiting for a design change or a funding confirmation. These sites are not failures. They are the program admitting a target was ambitious. The market reads them correctly: a stalled concept is not a canceled one, but its completion date is now written in conviction, not in contract. The engineer in Riyadh has watched two such sites near his own. “You learn not to read a paused site as dead,” he said. “You read it as waiting. The question is always what it’s waiting for — a decision, a partner, a permit. Some wait a season. Some wait years.” A second conversation, a year later I checked back with the engineer after his first assignment ended. His tower had opened, months late, and filled within a quarter. His read had sharpened. “If I could tell a newcomer one thing,” he said, “it’s this: bid the function, not the flag. The work tied to something the country already needs gets finished. The work tied to a dream gets finished when the dream pays its own way.” He had stopped chasing the most spectacular briefs. He now looks for assignments with a committed user on the other end — a ministry moving in, a line commuters will ride, a facility with a date it cannot miss. That discipline, more than any rendering, is how you survive the gap between the announced kingdom and the built one. The numbers behind the noise Stepping back from the individual sites, the aggregate tells a steadier story than the headlines. The construction program is large enough that even a portion of it completing represents more building than most regions see in a decade. The risk is not that the boom fizzles. It is that the boom delivers a different mix than the renderings promised — more metros and districts, fewer speculative wonders, and a pace set by physics rather than by ambition. That is the realistic read. The skyline is real. The certainty about which pieces arrive on time is not. The quiet work that finishes on time While the giga-projects grab attention, the kingdom’s steadier building — hospitals, schools, housing, utilities — tends to open closer to plan because it is smaller, less experimental, and tied to needs that do not depend on a future traveler. A new hospital serves a population that already exists. A power line serves commuters who are already there. These projects lack the drama of a car-free city, but they are the ones a contractor can bid with confidence, because the function on the other end is settled on the day the contract is signed. The lesson for an operator is to balance the portfolio. The marquee work wins headlines and relationships. The quiet work wins completion and cash flow. The firms that last in this market do both, and they use the quiet work to fund the patience the marquee work demands. The projects that will actually finish The headline number will keep growing, and the cranes will keep rising, because the program is real and the commitment behind it is deep. But the investor or contractor who treats the announced total as a settled pipeline will keep getting surprised by which sites open and which quietly slide. Read the function behind each project, not the rendering in front of it. The projects that will actually finish are the ones the country needs when the cameras leave, and that list is shorter, and more investable, than the list with the biggest numbers. The skyline sells the ambition. The completion list sells the reality. Buy the second, and you will not be the one explaining to a board why a flagship slipped a year.

SeaPRwire – A sourcing director I worked with spent the first half of last year moving part of his assembly line out of southern China. The arithmetic was convincing. Vietnamese factory wages sat around a third of what he paid in Dongguan. The workforce was young and large. A free trade agreement with the European Union meant his goods could land in Rotterdam without the tariffs he now faced. He signed a lease near Hai Phong, hired a local manager, and shipped his first containers in May. By February the delivery dates were slipping by three weeks at a time, and the cause was not on any spreadsheet he had built. This is the Vietnam story that the headline “companies are leaving China for Vietnam” leaves out. The relocation is real, and so is the bottleneck that follows it. The constraint is not tariffs or trade policy. It is the unglamorous machinery of moving things and training people inside a country that is growing faster than its own plumbing. The myth that Vietnam is just cheap China The easiest way to misunderstand Vietnam is to treat it as a lower-cost copy of the southern China manufacturing belt. That framing leads foreign teams to import a playbook that does not fit. China built its export engine over three decades, with highways, ports, power plants, and a deep bench of mid-level factory managers who could run a line without a foreigner in the room. Vietnam is compressing that timeline into a fraction of the time, and the seams show. The country did not start from the same base. Its road network connects the major hubs, but the last-mile links between an inland factory and a deep-water port are where time disappears. Its power supply has improved, yet industrial parks still schedule around peak-load limits. Its labor pool is huge, but the layer of supervisors and engineers who keep quality steady is thin and gets hired away as fast as it is trained. None of this means Vietnam is a bad bet. It means the bet pays off on a different schedule than the China playbook assumes, and the gap between the two schedules is the bottleneck. A useful way to frame the decision: Vietnam is not a cheaper version of China. It is a different manufacturing country at a different stage, and the savings are real only if you build the parts of the system China already had and Vietnam does not yet. The teams that treat the move as a copy-paste pay the bottleneck in surprise, through delays and defects they did not model. The teams that treat it as a build pay it in plan, through training budgets and second sources they chose on purpose. Both pay. Only one pays on terms the company controls. A small experiment that went sideways To see the bottleneck up close, watch what happens when a manager tries to move one container of finished goods from a factory in Bac Ninh to a customer in Europe. The factory finishes the order on time. Then the clock starts on everything the spreadsheet ignored. The manager books a truck to Hai Phong port. The road is fine until it isn’t; a stretch of highway widening means a two-hour detour. At the port, the container joins a queue that the booking system did not show. Customs clearance needs a document that the factory’s junior clerk formatted differently from the last shipment, and the discrepancy triggers a manual review. The review clears, but the vessel cutoff has passed, so the container waits for the next sailing, five days out. Multiply that by hundreds of containers a month and you see the shape of the constraint. It is not one dramatic failure. It is a dozen small frictions, each modest, stacking into weeks. The sourcing director’s three-week slippage was not bad luck. It was the system operating as designed, and the design was built for a smaller volume than the China-plus-one wave is now pushing through it. The north and the south are not the same country Vietnam’s manufacturing splits along a thousand-kilometer line, and the bottlenecks differ by region. Treating the country as one location is a mistake that costs teams time. The north, anchored by Hanoi and Hai Phong, is where most of the new electronics and furniture investment landed. It sits closer to China’s supply base, which shortens the import of components. But its ports and roads feel the pressure most acutely because so much new capacity poured in at once. Congestion is the northern signature. The south, anchored by Ho Chi Minh City and the ports of Cai Mep, carries the older garment and footwear base plus a growing electronics cluster. Its logistics backbone is more mature, but land and labor costs have climbed, and the distance from the Chinese border adds a day or two to inbound components. The southern signature is cost creep more than congestion. A team choosing a site should pick based on which bottleneck it can tolerate, not on which city has the cheaper lease. The north saves you on the supply chain but tests your patience at the port. The south smooths the logistics but squeezes the margin. A concrete example makes the split clear. A furniture exporter shipping through Cai Mep can usually get a container loaded with less queue time than its peer in the north, but it pays more for land and watches component trucks take an extra day from the Chinese border. The northern peer saves that day on inbound parts and sits closer to the electronics supply cluster, yet budgets an extra week into every Europe-bound shipment for port congestion. Neither site is wrong. Each simply trades one bottleneck for another, and the right pick depends on whether your business dies from delay or from margin. What a failed relocation teaches One apparel brand tried to move a quarter of its production from China to a new line near Ho Chi Minh City. The lease was cheap, the workers were available, and the first samples passed inspection. Then the problems that no spreadsheet captured arrived. The line could not hold consistent stitch quality because the supervisors who knew the specs had been poached by a neighboring factory two weeks after training. The local fabric supplier delivered late three times in the first four months, and there was no second source within driving distance. The quality team flagged defects that the new workers had not been trained to catch. By month six the brand was air-freighting a fifth of its output to hit deadlines, wiping out the wage savings. The lesson is not that Vietnam failed the brand. The lesson is that the brand imported a China assumption: that supervisors, suppliers, and trained hands would be there on day one. They were not, because Vietnam’s middle layer is the exact part of the system under the most strain. The relocation succeeded on paper and struggled in practice because the bottleneck was human, not physical. The broker and the buffer Two mitigations show up again and again among teams that tame the bottleneck. The first is a customs broker who has handled your exact product category, not a generic freight forwarder. The second is an inventory buffer parked at the right point in the chain. A broker who knows your codes and your port’s habits turns the document-tax from a weekly surprise into a routine. The buffer is the harder call. Holding an extra week of components near the factory costs carrying capital, but it absorbs the delay that would otherwise stop the line. Teams that refuse any buffer because it looks wasteful are the ones that halt production when a single late shipment arrives. The teams that last treat the buffer as insurance they hope not to use, not as waste they refuse to fund. A category-specific broker beats a general forwarder for cutting clearance delays. A small component buffer near the line prevents stoppages that cost more than the buffer. The mitigations that matter are boring and local, not strategic and grand. The real constraint is a people problem Strip away the roads and the ports and the power, and the bottleneck that limits Vietnam is the thin layer of trained mid-level talent. Factories can be built in months. A supervisor who can hold a line to spec, a planner who can see a bottleneck before it forms, an engineer who can tune a machine without calling the vendor — those take years to grow, and Vietnam is growing them slower than the factories are arriving. The numbers tell the story in plain terms. A line worker in a Vietnamese factory might earn the equivalent of roughly 200 to 250 US dollars a month, against perhaps 600 to 800 in the Chinese coastal belt she is meant to replace. That gap is the headline. But the supervisor who keeps that line honest earns far more, and there are not enough of her to go around. Foreign-invested plants bid up the limited pool, which pushes smaller domestic suppliers out of reach of the people they need. The wage gap draws the investment, but the supervisor gap limits what the investment can actually produce. Training programs exist, yet they lose graduates to higher bidders faster than they create them. The bottleneck moves up the skill ladder: it is no longer about hands, it is about the people who lead hands. Pricing the move in real terms For a finance team, the Vietnam case often looks like a straight wage arbitrage. It is not. The true cost sheet has lines the original model omitted. Start with the visible saving. If a line that cost 800 dollars a worker in China runs for 230 in Vietnam, the labor line drops by more than half. Then add the hidden lines. Air freight to cover delays. A second quality audit layer the China plant did not need. Consultants to build the supervisor pipeline the local market could not supply. A longer cash cycle because the logistics chain is slower and the financing follows the slower clock. A procurement lead who tracked this for a furniture exporter told me the wage saving held, but the total landed cost advantage shrank from the pitch-deck 35 percent to something closer to 12 percent once the frictions were counted. That is still a win. It is just a smaller, more honest win than the relocation story implied. The customs and paperwork tax Beyond physical movement, Vietnam’s border procedures add a tax of their own in time. Customs clearance is generally competent, but it rewards precision. A misclassified item, a mismatched invoice, a certificate formatted a hair off — any of these can pull a shipment into manual review. For a team new to the market, those reviews are frequent at first and taper as the local clerks learn the rhythm. The practical fix is unglamorous. Hire a customs broker who has handled your product category, build the document templates once and lock them, and train the factory clerks to the template rather than letting each shipment be a fresh attempt. Teams that skip this step pay the time tax every week. Teams that invest in it recover most of the delay within a quarter. Where the smart money is going The companies getting the most from Vietnam are not the ones chasing the cheapest lease. They are the ones building the missing layer. They co-invest in training their own supervisors instead of poaching. They dual-source components so a late delivery does not stop the line. They locate near a port that fits their volume rather than a city that fits a brochure. A few moves separate the winners from the rest: Build the supervisor bench on your own dime; do not assume the market will hand it to you. Keep a second source for anything that can halt production if it is late. Choose the site around the bottleneck you can live with, not the one with the nicest incentives. Budget for the time tax in the first year; it is real and it is temporary if you work it. The power question nobody asks Vietnam’s electricity story is better than a decade ago but still a live risk. Industrial parks allocate power, and during peak summer some parks ask factories to shift heavy loads to off-peak hours. For a line that must run continuously, that means rescheduling and sometimes losing output. A team I spoke with lost two production days in a single hot month to load-shedding coordinated by the park. The country is building capacity fast, including imported gas and new plants, but the gap between demand and supply tightens exactly when the China-plus-one factories switch on. This is a different class of bottleneck from the port queue. You can plan around a port by booking earlier. You cannot easily plan around a grid that tells you to stop. The practical mitigations are real but costly: on-site generators, a battery buffer, or locating in a park with a secured supply. Each adds to the landed cost that the wage arbitrage was supposed to deliver, and each is a line the original relocation model rarely included. The teams that price it honestly are the ones that survive the first year. Why the relocation wave made it worse The bottleneck is not only Vietnam’s. It is Vietnam plus a historic surge of companies arriving at once. The China-plus-one strategy went from a hedge to a mandate for many boards after supply chains seized during the pandemic and trade tensions rose. When hundreds of foreign factories converge on the same handful of industrial parks in the same two years, the local system that was adequate for gradual growth gets hit by a wave. The ports were sized for the old volume. The supervisor pool was sized for the old volume. The customs clerks were sized for the old volume. All three are now processing far more than they were built for, and the queues are the visible symptom. Vietnam is responding with new ports, new parks, and training pushes, but infrastructure lags demand by its nature. The bottleneck is therefore temporary in the long run and acute in the short run, and most of the pain is being felt right now. A relocation done right Contrast the struggling apparel brand with an electronics assembler that arrived with a different plan. This team spent its first six months building the layer before building the volume. It opened a training center with a local technical school, guaranteed jobs to graduates, and paid above market to keep them. It dual-sourced its three critical components from day one. It chose a northern park specifically for port access even though the lease cost more. By month twelve the assembler’s defect rate matched its China line, and its on-time delivery held. The wage saving was real and the hidden costs were smaller because the team had pre-built the missing pieces. The difference between the two relocations was not luck or location. It was whether the company treated the bottleneck as a problem to solve or a surprise to absorb. The small supplier squeeze One more pressure deserves naming. As foreign factories bid up supervisors and engineers, the domestic Vietnamese suppliers that feed them get squeezed out of the talent they need. A local packaging shop cannot match the salary a top-tier plant offers a quality lead. So the supplier’s own quality slips, which then forces the foreign plant to inspect more, which raises costs again. The bottleneck propagates backward through the supply chain, and the companies that overlook their suppliers’ strain inherit it. The wage gap that drew the investment also starves the local suppliers of the people they need. A supplier’s slipped quality becomes the foreign plant’s added inspection cost. The fix is to develop suppliers, not just audit them after they fail. Pricing the inland leg in real terms A container that costs, say, 1,200 dollars to move from Shenzhen to Rotterdam might cost around 1,050 from Hai Phong once the freight-rate difference is counted. But the inland leg from a Bac Ninh factory to Hai Phong can add 80 to 120 dollars in trucking, plus the risk cost of delay. Expressed in dong, that inland leg alone runs into the tens of millions, a sum a small exporter feels in working capital. The point is not the exact figure. It is that the saving on the ocean leg gets partly eaten on the road and at the port, and the eating is variable, not predictable. What to watch by year three Teams that survive year one and year two face a different test in year three. The wage gap that drew them begins to close as local costs rise and the currency’s stability is tested by imported inflation. The incentive packages that sweetened the lease expire. The easy efficiency gains from simply organizing a chaotic line are spent. What remains is whether the company built durable local capability or just rode the arbitrage. The ones that last are usually the ones that localized real decision-making, developed suppliers they do not have to babysit, and trained enough of their own people to stop depending on poaching. The ones that leave are the ones that treated Vietnam as a temporary wage play and found the temporary part was shorter than the gap they were filling. Vietnam’s real bottleneck is a people problem The relocation from China is genuine, and Vietnam will keep taking share. But the constraint that decides which companies actually win here is not tariffs, ports, or power. It is the rate at which a thin layer of trained human talent can be grown to match the factories arriving faster than the training system can feed them. The road projects will finish. The ports will deepen. The power will stabilize. The supervisor gap is the one that will keep biting long after the concrete sets, and the teams that plan for it are the ones that will still be here in five years.

SeaPRwire – A founder I know sat in a co-working space near Tanjong Pagar last spring, staring at a spreadsheet that would decide where his company’s regional headquarters would live for the next five years. His startup sold inventory software to mid-sized retailers across Asia. The spreadsheet had three columns: Singapore, Kuala Lumpur, and Jakarta. The salary line for a finance lead was S$9,800 in Singapore, RM11,000 in Kuala Lumpur, and Rp48 million in Jakarta. Rent for the office he wanted ran S$14 per square foot per month in Singapore, less than half that in the other two. He picked Singapore anyway. Not because it was cheap. It was the most expensive option on every line. He picked it because, as he put it later, “when a customer in Manila or a bank in Bangkok asks where we’re headquartered, the answer opens doors or closes them.” That sentence captures the whole Singapore regional-HQ story better than any ranking does. The door-opening effect is real, not vanity Companies do not move their regional headquarters to Singapore for the weather or the skyline. They do it because a Singapore address still signals something to counterparties who will never visit the office. A payments firm raising a Series B, a logistics company bidding for a government tender in the Gulf, a beauty brand signing a distributor in Vietnam — all of them find that the incorporation certificate with “Singapore” on it shortens conversations that would otherwise drag for months. This is not magic. It is a pile of boring, cumulative advantages that competitors cannot copy with a press release: A corporate tax rate of 17%, with generous partial exemptions that bring effective rates lower for most small and mid-sized firms. A network of more than 100 bilateral tax treaties, which matters enormously when you move money across borders and do not want to be taxed twice. English as the working language of courts, contracts, and board meetings. A timezone that lets you take a morning call with Sydney and an evening call with Bangalore without much pain. Talent that arrives on a work pass within weeks, not quarters. None of those items alone is decisive. Stacked together, they are why a city of under six million people hosts the regional headquarters of thousands of multinational firms, from semiconductor giants to consumer-app companies. What the cost sheet actually shows Let me return to that spreadsheet, because the numbers deserve a closer look. The founder’s finance lead cost S$9,800 a month in Singapore. In Kuala Lumpur the same role might cost RM9,000 to RM12,000. On paper, Malaysia wins. But the spreadsheet did not capture two things that showed up three months later. First, the Singapore hire had worked at two regional banks and already knew the compliance language of three regulators. The founder estimated that prior experience saved roughly four months of onboarding. Four months of a delayed product launch, in his words, “cost more than the salary gap for a year.” Second, the Singapore entity could open a corporate bank account with a regional bank in under two weeks. His friend who set up in another Southeast Asian hub waited eleven weeks for the same basic account, and the delay pushed back a payment-integration partnership by a full quarter. Here is a stripped-down version of the math he eventually shared with me, converted to round figures: Singapore setup: S$14,000 office deposit + S$9,800 monthly senior salary + S$2,100 monthly compliance retainer. Kuala Lumpur setup: S$9,000 office deposit + S$6,400 equivalent monthly senior salary + S$1,200 monthly compliance retainer. The gap: about S$5,300 a month in ongoing cost, or S$63,600 a year. He decided the S$63,600 was cheap insurance against the four-month launch delay and the eleven-week banking wait. That is the real decision framework. Not “which is cheapest,” but “which cost can I not afford to pay in time and trust.” Three companies, three very different reasons The regional-HQ choice looks different depending on what your business actually does. Three examples, none of them hypothetical in shape, show the range. A cross-border ecommerce operator moved its regional HQ to Singapore mainly for the treaties. It ships from warehouses in China and Vietnam to customers in six countries. Every one of those flows touches import rules, value-added tax, and repatriation of profit. A Singapore holding structure, paired with the right treaty network, let the company route dividends without a second tax bite in two of those markets. The founder said the savings paid the entire Singapore overhead in year one. A software-as-a-service firm came for the talent. Its product needs solution engineers who can speak to enterprise buyers in English, Mandarin, and Bahasa. That specific blend is easier to hire in Singapore than almost anywhere else in the region. The company keeps a large delivery team in the Philippines and a support hub in India, but the brain — pricing, partnerships, product strategy — sits in Singapore because that is where the people who can do all three at once actually live. A family-owned trading house relocated from Hong Kong for regulatory calm. It imports food ingredients and re-exports them across the region. The principals wanted a jurisdiction where the rules change slowly and the regulator answers the phone. Singapore fit. They kept a small Hong Kong desk for certain China-facing deals, but the center of gravity moved. Notice that none of these three picked Singapore for the same reason. That is the point. The city wins not by being best at one thing, but by being good enough at everything a regional headquarters needs. A conversation that changed one founder’s mind I sat in on a call between that inventory-software founder and a mentor who runs a fund. The mentor pushed back hard on the Singapore premium. Mentor: “You’re burning cash on rent you don’t need. Your customers are in the Philippines and Indonesia. Why pay Singapore prices?” Founder: “Because when I walk into a bank in Manila to negotiate a working-capital line, they Google my company. They see Singapore. The meeting lasts forty minutes instead of being rescheduled three times.” Mentor: “That’s a story. Show me the number.” Founder: “The number is the term sheet. Two banks declined us from a Jakarta entity last year. Same pitch, Singapore entity, one said yes. I can’t prove it’s the address. But I’m not repeating the experiment.” The mentor laughed and dropped it. The founder was right to hold his ground. The value of a regional HQ is rarely a single line item. It is the sum of a hundred small moments where trust is granted a little faster. The blind spots nobody puts in the pitch deck Singapore is not a free lunch, and the companies that struggle here make the same mistakes. The pitch decks never mention them. The first blind spot is the talent war at the junior level. Everyone wants the same bilingual, three-years-experience analyst. Salaries for that profile have climbed faster than inflation, and attrition is brutal. A regional HQ that assumes it can hire and keep such people cheaply will bleed. The second is the “everything is nearby” trap. Singapore is a hub, but it is still a small island. Your warehouse is not here. Your factory is not here. Your customers, mostly, are not here. The HQ exists to coordinate, not to produce. Founders who expect Singapore itself to generate revenue often leave disappointed. The third is compliance drift. The same stable regulator that attracts you will fine you for the same sloppy filings that another jurisdiction might ignore. A regional HQ that treats Singapore like a lightly-touched mailbox gets caught, and the penalties are public. Junior talent is expensive and restless; budget for replacement, not just hire. Revenue is generated elsewhere; the HQ’s job is coordination, not sales. The regulator is strict and the penalties are visible; treat filings as real work. The local view, from someone who places talent A recruiter I spoke with places regional finance and operations leaders for multinational clients. She has done this for eleven years. Her read is unsentimental. “Companies come for the brand and stay for the boring stuff,” she told me. “The brand gets them in the door. The treaties, the banks, the time zone, the schools for expat kids — that’s what makes them renew the lease.” She sees the same pattern repeatedly. A company arrives with two expatriate directors and a skeleton local team. Within eighteen months, if it is still here, the local team has grown and the expats have either left or shifted to regional travel roles. The ones that leave usually did so because they never built that local core. “You can’t run a regional HQ from a hotel room,” she said. “The ones who treat it as a postbox go home.” When Singapore is the wrong call It is worth saying plainly: a Singapore regional HQ is not always right. If your entire operation is a factory in one province of Vietnam, a Singapore HQ adds cost and distance with little benefit. If your customers are all in one large market and you have no cross-border money movement, a local entity serves you better. If your margin is thin and your staff count is small, the premium can swallow you. The decision should follow the money flows, not the prestige. Ask three questions: Do I move money across borders? Do I need to signal stability to partners I have never met? Do I need trilingual senior talent that is hard to find elsewhere? If the answer is yes to two of those, Singapore probably earns its premium. If not, the rent is just rent. What the alternatives quietly cost you The spreadsheet comparison almost always pits Singapore against its neighbors on rent and salary. That is the visible part. The hidden part is what each alternative takes back in friction. Kuala Lumpur is the closest substitute. Offices cost less than half, and trilingual talent is genuinely available. The catch is the banking and the treaties. A Malaysian entity can still open regional accounts, but the treaty network is narrower, and some Gulf and European counterparties treat a Kuala Lumpur address with more questions than a Singapore one. For a company that mostly serves Malaysia and Indonesia, the trade is fine. For one raising foreign capital or moving dividends across four markets, the savings erode. Jakarta offers cheap labor and a vast domestic market, but the regional-HQ case is weakest. The language of regional business is still English in most boardrooms, and the permitting timeline for certain licenses stretches. A firm that wants Jakarta as a base for Indonesia, not as a springboard to the region, is making the right call. A firm that wants it to coordinate Southeast Asia usually regrets it within two years. Bangkok sits between the two. Strong tourism and consumer story, decent talent, improving infrastructure. But the regional headquarters community is thinner, and the legal certainty that Singapore sells is not yet where Bangkok needs it to be for risk-averse treasuries. The pattern is consistent. Every alternative saves you money on the lines you can see and charges you on the lines you cannot: The neighbor that saves you rent usually charges you in banking delay. The neighbor with cheaper talent often lacks the treaty cover for your money flows. The neighbor with a huge home market may not speak the region’s boardroom language. How a tax treaty actually changes the math A treaty sounds abstract until you see the wire. Suppose the Singapore holding company owns the Vietnam operating unit. The Vietnam unit earns profit and wants to send a dividend to the parent. Without a treaty, Vietnam withholds tax at its domestic rate on that dividend, often a steep one. With the Singapore-Vietnam treaty, the withholding rate drops to a much lower figure, sometimes single digits. Multiply that across years and across units, and the saving is not rounding error. The same logic applies to royalties and interest. A regional headquarters that licenses software to its own subsidiaries can route those payments through Singapore and pay far less withholding than it would if the payments bounced through a jurisdiction without the right treaties. This is why the cross-border ecommerce operator in the earlier example said the treaties alone paid for the overhead. The tax rate gets the headlines. The treaty network does the heavy lifting. One more case: the fintech A payments fintech moved its regional headquarters to Singapore for a reason none of the others cited: the regulator. It needed a license to move money across borders, and it needed that license to be respected by partner banks in eight markets. Singapore’s licensing framework is recognized, and a Singapore-licensed money institution gets a hearing that a lesser-known jurisdiction does not. The founder estimated the licensing credibility shortened partner onboarding by roughly half a year. For a payments company, half a year of partner access is the difference between a product that launches and one that stalls in compliance limbo. Standing up the HQ, step by step For a founder who has decided Singapore is right, the actual setup is less mysterious than the pitch decks suggest. The outline looks like this. Pick the entity type. Most regional headquarters use a private limited company; some use a holding company if the structure is purely about owning subsidiaries. Reserve the name and incorporate. This can finish in a day or two once documents are ready, but bank-ready paperwork takes longer than the incorporation itself. Open the corporate bank account. This is the step that surprises people. Budget two to four weeks, line up the directors’ documents early, and expect questions about the source of funds. Apply for the employment passes. The senior hires who justify the HQ usually qualify, but the paperwork rewards precision, not optimism. Set the compliance calendar. The filings that keep the entity clean are quarterly and annual; miss them and the penalties are public. None of those steps is hard in isolation. The discipline is doing all five without letting the bank account become the bottleneck that stalls everything behind it. The wealth angle nobody puts on the slide Singapore also became the regional home for family offices, and that feeds the headquarters ecosystem in ways the cost sheets miss. A family office that relocates brings discretionary capital, and that capital backs the regional headquarters of operating companies through convertible notes, venture rounds, and trade finance. The founder near Tanjong Pagar did not raise from a family office, but two of his peers did, and both said the same thing: the meeting happened because they were already in town. This is the part of the Singapore story that does not show up in any spreadsheet. The density of capital, operators, and regulators in a three-kilometer radius means a chance encounter at a lunch can become a term sheet. You cannot expense serendipity. But you can put yourself where it happens, and that is a large part of why the regional headquarters keeps renewing its lease. The second-year trap Most regional headquarters that fail in Singapore do not fail in year one. They fail in year two, after the expat directors go home and the local team is not ready to carry the load. The fix is unglamorous: hire the local core in month three, not month fifteen. A few habits separate the ones that last from the ones that quietly close: Put a local person in a real decision role early, not as a deputy waiting for instructions. Treat the entity as a living operation with filings, not a mailbox with a plaque. Spend the first year building regulator and banker relationships face to face, because those relationships are the actual product. The signals beyond the certificate A Singapore address carries softer signals that never appear on a cost sheet but shape real decisions. A regional manager relocating with school-age children looks at where the international schools rank, and Singapore’s cluster is consistently strong. A sales lead who flies out every Monday looks at Changi’s connections, and few hubs in the region match the same-city reach to secondary cities. A general counsel looks at where commercial disputes get resolved, and Singapore’s arbitration reputation is the reason many cross-border contracts name it as the venue before a word of negotiation happens. These are not luxuries. They are retention and credibility tools: Strong schools keep the people you relocated from asking to go home after a year. Flight reach turns a three-leg trip into a direct one, and time is the one input you cannot buy more of. A trusted arbitration venue lets you sign contracts with strangers faster, because both sides already know where a fight would land. The certificate opens the door. These signals keep it open. None of this is unique to Singapore in principle. What is unusual is how many of these signals line up in one small place at the same time. That alignment is the real product being sold, and the rent is the price of standing inside it. The tax break is the least of it People fixate on the 17% corporate tax rate as if that were the whole story. It is not even the most important part. The treaties, the banking speed, the talent blend, the regulatory predictability, the timezone — those are what actually move the needle for a regional headquarters. The tax number is the line item everyone understands, so everyone talks about it. A regional HQ is a trust machine. You build it so that strangers across a dozen markets will take your call, sign your contract, and lend you money before they have met you. Singapore happens to be the most reliable factory for that particular machine in this part of the world. The founder near Tanjong Pagar understood that before he finished his spreadsheet. The companies that do best here are the ones that figured it out before they arrived, not the ones hoping the address would do the work for them.

Indonesia – October 21, 2026 – (AsiaPressWire) – Indonesia banned the export of unprocessed nickel ore in January 2020, pulling forward a deadline that had originally been set for 2022. Five years on, it is the most consequential industrial policy experiment in Southeast Asia, and the clearest test anywhere of whether a commodity-producing country can force its way up the value chain by legislating against its own exports. The answer is genuinely mixed, and the mix is more instructive than either the celebratory or the dismissive version of the story. Indonesia did build a processing industry that did not exist. It also created chronic global oversupply, locked itself into a technology and capital structure it does not control, and discovered that banning an export is much easier than capturing the margin that sits at the end of the chain. This piece is about what the policy actually built, what it cost, where the money actually goes, and what a company deciding whether to commit capital there needs to test before it writes a cheque. The policy, stated precisely The measure most people mean by “downstreaming” is a prohibition on exporting nickel ore, introduced through a mineral and coal mining regulation and brought forward to take effect in January 2020. The logic was straightforward: if ore cannot leave, processors must come to the ore. Three things about it are worth being precise about, because the shorthand versions mislead. It is not a tax. It is a prohibition. That distinction matters for modelling, because a tax can be priced and hedged while a prohibition reorders the physical geography of an industry. It is not limited to nickel. Indonesia has applied the same logic to bauxite and has extended the approach, in various forms, to other commodities including copper and tin. Companies treating this as a nickel-specific rule are reading the policy too narrowly. It is not static. Thresholds, permitted product forms, royalty structures, and domestic-market obligations have been adjusted repeatedly since 2020. Any model built on the rules as they stood in a given year needs re-checking against the rules as they stand now. The instrument has also been paired with inducements: tax holidays, investment allowances, and the designation of industrial estates where processing is concentrated. The two largest are the Morowali industrial park on Sulawesi and the Weda Bay complex on Halmahera in North Maluku. Most of the country’s processing capacity sits inside those two sites, which is itself a concentration risk worth naming. What the country actually became On the narrow question of whether the policy worked as industrial policy — did processing capacity appear? — the answer is unambiguously yes. Indonesia moved from exporting raw ore to exporting ferronickel, nickel pig iron, matte, mixed hydroxide precipitate, and, further along, nickel sulfate and precursor material. By the middle of the decade it accounted for a share of global mined nickel supply approaching or exceeding half, depending on the year and the source consulted, up from a fraction of that before the ban. Export revenue from nickel products rose from low single-digit billions of US dollars in the ore era to tens of billions afterwards — a genuine, large, and fast shift in the composition of the country’s exports. Employment at the two main parks runs to tens of thousands of direct jobs. That is a real number in a country where the alternative use of remote land in Sulawesi and Halmahera was close to nil. What is less often stated precisely is what that capacity is for. The overwhelming majority of Indonesian processing capacity was built for stainless steel — ferronickel and nickel pig iron feeding the stainless chain — rather than for batteries. The battery-grade portion, produced mostly through high-pressure acid leaching into mixed hydroxide precipitate and then into sulfate, is real and growing but remains the smaller part of the complex. A country that banned ore exports to serve the electric vehicle revolution ended up, in its first phase, mostly serving stainless steel. Where the chain actually stops StageWhat it producesWhere Indonesia sitsWhat it capturesMiningLaterite and saprolite oreDominant global positionLowest margin, highest physical footprintSmelting (RKEF)Ferronickel, nickel pig ironVery large, largely builtCommodity margin, tied to stainless demandRefining (HPAL)Mixed hydroxide precipitateGrowing, capital-intensiveHigher margin, exposed to cobalt and acid costsSulfate and precursorBattery-grade nickel chemicalsPresent but limitedApproaching cell-grade pricingCathode and cellCathode active material, cellsEarly / under constructionThe margin everyone wantsPack and vehicleBattery packs, EVsAssembly onlyThe margin downstreaming was aimed at The table is the argument. Indonesia is unambiguously dominant at row one, very strong at row two, present at row three, limited at row four, and not yet established at rows five and six. The stated ambition of the policy was row six. The delivered result, five years in, is mostly rows one to three. This is not a criticism of execution so much as a description of what happened. Each step up the table requires a different input — not more ore, but technology, process know-how, qualification with cell makers, and access to a customer who will accept your material into a battery supply chain. Those are not things a prohibition on ore exports can compel. They have to be bought, licensed, or built over years. The stainless anchor, and why it still sets the floor Battery demand gets the attention, but stainless steel still absorbs the majority of primary nickel, and that fact is what makes the Indonesian build-out financially legible at all. RKEF lines producing ferronickel and nickel pig iron have a customer base that exists today, in volume, and does not require qualification. That is why the first phase of downstreaming went there rather than into battery chemicals: the capital was available, the technology was mature, the off-taker was known, and the product could be sold the day the line started. This has a stabilising effect that is easy to miss. When battery-grade nickel prices weaken or when a cell maker delays a programme, the stainless chain keeps absorbing metal. It acts as a floor under offtake, and therefore under utilisation, in a way that a purely battery-facing industry would not have. It also means that the economics of most Indonesian capacity are driven first by stainless demand in China and India and only second by electric vehicle forecasts — which is the opposite of the framing most coverage uses. The consequence for anyone modelling this market is that “nickel demand” is not one number. It is two largely separate demand pools with different price sensitivities, different customers, and different contract structures, connected mainly by the fact that producers can switch some capacity between them. Treating them as a single market is why so many forecasts of the 2021–2023 period were wrong in the same direction. What it cost The costs are real and they fall in four places. Capital. Smelting and refining are among the most capital-intensive industrial activities in the commodity world. A rotary kiln–electric furnace line is expensive; a high-pressure acid leach plant is more expensive still, and its economics depend on ore grade, acid supply, and the cobalt credit in the byproduct stream. The billions of dollars invested came overwhelmingly from foreign capital, and the single largest source of that capital is Chinese. Power. Indonesian smelting is heavily coal-fired, with captive power plants built alongside the lines. That has two consequences. The carbon intensity of the resulting nickel is high, which increasingly matters for buyers who have to report it. And it ties the economics of the metal to the cost and availability of domestic coal, which the government also prices and regulates. Environment. Laterite mining clears land and moves very large volumes of material. The tailings question is the hardest one, and in parts of the country the practice of disposing of tailings at sea has drawn sustained criticism. Acid leaching creates its own waste stream. Companies sourcing Indonesian nickel and reporting against European battery or supply-chain due diligence rules should assume these issues will be raised, and should have sourcing documentation that withstands it. Market structure. This is the cost that was least anticipated. Indonesia added processing capacity faster than global nickel demand grew. The result, visible through 2023 and into the following years, was sustained oversupply and a nickel price that fell well below the level at which the newest and highest-cost capacity is comfortable. Parts of the industry outside Indonesia — Australian laterite operations and several projects elsewhere — were squeezed or shut. Then the pressure came home: margins inside Indonesia compressed as well. The part the brochures skip: who captures the margin A prohibition on ore exports guarantees that processing happens in the country. It does not guarantee that the profit from processing accrues to the country or to its companies. The ownership structure of much of the built capacity is foreign. Offtake is frequently tied to the same groups that provided the capital. Financing, engineering procurement, and construction were largely supplied from outside. In that arrangement, Indonesia receives employment, taxes and royalties, the depreciation of a large physical asset base, and the strategic position of being indispensable to a critical supply chain. Those are not nothing — employment and fiscal receipts are the point of industrial policy — but they are not the same as capturing the margin at the cell and vehicle end of the table. There is a second-order effect here that deserves attention. Because the dominant capital and offtake relationships run through Chinese groups, Indonesian nickel has become entangled in the classification rules of other countries’ industrial policy. The United States clean-vehicle credit regime applies foreign-entity-of-concern restrictions that trace ownership and control through the supply chain. European rules apply different but overlapping tests. The practical consequence is that material which is perfectly saleable into one market may be disqualified from another, and the disqualification can be triggered by an ownership stake several levels up rather than by anything about the metal itself. For a company building a supply chain, this means the question is never simply “is there nickel available in Indonesia.” It is “which corporate entity produced it, who controls that entity, and where does the resulting material qualify.” Qualification: the step nobody budgets for Selling metal is one thing. Selling metal into a battery supply chain is a different and much slower business, and the gap between the two is where a great deal of announced capacity quietly stalls. Battery makers do not buy material on specification sheets. They qualify it, plant by plant, through a process that runs from months to well over a year. Samples are produced, cell-level testing is run against cycle life and safety behaviour, the plant’s process control is audited, and the material is then approved against a specific production line. Change the plant, change the process, or in some cases change the ore blend feeding it, and the qualification has to be revisited. This has three practical implications. First, a refinery that has not been qualified has no battery customer regardless of how good its product is on paper; it can only sell at commodity pricing, which is precisely the margin it was built to escape. Second, the bottleneck is not always chemistry — it is often simply the queue at the cell makers, who can only run a limited number of qualification programmes at once. Third, the value of an existing qualification is high and underappreciated in deal valuations, because it cannot be bought quickly at any price. There is a related trap on the buying side. A purchaser that signs an offtake for “battery-grade nickel” from a plant under construction is not buying battery-grade nickel; it is buying a schedule, a qualification process, and a set of risks that sit between the two. Contracts that do not distinguish those states — and that do not allocate the cost of a failed or delayed qualification — are the ones that end up in dispute. What other countries tried, and what happened Indonesia is not the first country to try this, and the comparison is instructive because the results differ in ways that track the specifics rather than the ambition. Chile has moved on lithium through a different instrument — a state-participation model for its salt flat operations, with the state taking control or majority stakes in new projects rather than a blanket export prohibition. The approach preserves more room for existing operators and leaves the question of where processing happens partly open. It also moves more slowly. Guinea holds the world’s largest bauxite reserves and has pushed for domestic alumina refining for years, with results that have been slower and more uneven than Indonesia’s, largely because the capital and infrastructure requirements are less forgiving and the bauxite market is less concentrated. Zimbabwe and Namibia both moved to restrict raw lithium exports, with Zimbabwe banning unprocessed lithium ore exports. The stated logic is identical to Indonesia’s. The scale of the resulting domestic industry, so far, is not — which is a reminder that a prohibition only works as industrial policy when there is capital ready to build against it and a customer willing to buy the output. The Philippines is the closest analogue and the most interesting case. It is a large nickel ore producer and has debated an export ban for years without implementing one, which has left it exporting ore into the Indonesian complex it might otherwise have competed with. The comparison suggests that the speed and decisiveness of Indonesia’s move mattered at least as much as the move itself. Across all of these, the pattern is consistent. Prohibitions work when three conditions hold: the country’s share of supply is large enough that the ban changes global trade rather than just diverting it, capital is available and willing to build on the country’s terms, and there is a market for the processed product that does not require the producer to win a qualification race first. Indonesia had all three for stainless-grade material and only the first two for battery-grade. Questions companies actually ask Can we get battery-grade material, or only stainless-grade? Both exist, but battery-grade capacity is the smaller share and is concentrated in specific plants. Qualification with a cell maker is a separate and slower process from buying metal. Is the ore export ban at risk of being reversed? Treat it as permanent policy direction, not a temporary measure. It has survived changes of administration and legal challenge. Model it as a structural feature of the market. What happens if the price stays low? Capacity built for a higher price does not disappear; it runs at lower margin, and consolidation follows. For a buyer, low prices are favourable. For an investor, they change the project’s whole return profile, and the projects most exposed are the newest and highest-cost ones. How do we handle the carbon intensity? You report it. The footprint of coal-fired smelting is materially higher than alternative supply, and buyers subject to reporting obligations should be sourcing data at the plant level rather than the country level. Does Indonesian material qualify for incentives in our target market? Check ownership and control, not just origin. This is the single most common and most expensive sourcing error in this market right now. Three ways this resolves The next five years are genuinely uncertain, and the honest framing is scenario rather than forecast. Consolidation. Prices stay low enough that marginal capacity stops running, and the industry concentrates into the lowest-cost operators. Indonesia’s position strengthens on volume while its fiscal take per tonne weakens. This is, on present evidence, the path of least resistance. Escalation. Indonesia moves decisively up the table — cathode, cell, and eventually vehicle assembly, supported by incentives and by requirements attached to domestic market access. This requires technology partnerships and customer qualification that no export rule can mandate. Several projects are aimed at it. Whether they reach commercial scale on schedule is open. Fragmentation. Buyers in the United States and Europe build qualified supply outside Chinese-controlled chains, while Chinese-linked capacity serves the rest of the market. Indonesian volume keeps growing but splits into two pools with different prices and different customers. Elements of this are already visible. These are not mutually exclusive. Consolidation and fragmentation can happen at the same time, to different parts of the industry. Glossary Laterite — The ore type Indonesia mines, requiring different processing from the sulphide ores common elsewhere. RKEF — Rotary kiln–electric furnace, the route to ferronickel and nickel pig iron for stainless steel. HPAL — High-pressure acid leaching, the route to mixed hydroxide precipitate for battery-grade material. MHP — Mixed hydroxide precipitate, the intermediate product that HPAL produces and refiners convert to sulfate. FEOC — Foreign entity of concern, the ownership-and-control test that determines clean-vehicle credit eligibility in the United States. The risks worth pricing Technology lock-in. Much of the built capacity was designed around the ore, the capital, and the customer relationships available at the time it was built. If battery chemistry shifts away from high-nickel formulations, that capacity is less valuable than its construction cost. Environmental liability. Land disturbance and tailings practice are live issues now and will remain so. Expect tightening, and expect the compliance cost to land on producers and then on buyers’ documentation requirements. Concentration. Two industrial parks carry most of the capacity. A single-site disruption — power, labour, logistics, regulation — is a supply-chain event, not a local one. Policy drift. Royalty rates, export levies, domestic processing obligations, and local content requirements have all moved. A model built today should be stress-tested against at least two alternative policy settings. The commodity cycle itself. Nickel is a cyclical commodity with a volatile price history, including a short squeeze in 2022 that remains the clearest illustration of how unstable this market can be. Any project economics built on a spot price rather than a through-cycle assumption should be redone. A checklist before committing capital Establish which row of the chain you are actually investing in, and whether the margin at that row justifies the capital Trace ownership and control of your counterparty to the level required by your target market’s rules — not to the level that is convenient Source emissions data at plant level, and confirm it survives your reporting obligations Model at least two nickel prices and two policy settings; if the project only works at one, it is not a project Confirm power supply, logistics, and port access are contracted rather than assumed Check whether your offtake customer has already qualified the specific plant, or whether that process has not started Ask what happens to your asset if battery chemistry moves away from high-nickel The honest bottom line Indonesia’s nickel policy is neither the triumph its supporters describe nor the cautionary tale its critics prefer. It did the thing it set out to do: it moved processing into the country, created a large industrial base where there was essentially none, and changed the structure of a global commodity market within five years. Very few industrial policies in the developing world have done that much, that fast. What it did not do is the harder part. Banning an export moves the location of an activity. It does not move the location of the margin, the technology, or the customer relationship — those are earned at the top of the chain, and they are earned by qualification, process capability, and access, none of which respond to a prohibition. For a company deciding what to do about it, the practical conclusion is unromantic. Indonesian nickel is abundant, cheap, and increasingly central to the world’s battery supply. It is also concentrated, carbon-intensive, policy-exposed, and tangled in other countries’ ownership rules. Those facts are not contradictory. They are the deal, and the companies that do well here are the ones that priced all of them rather than the ones that found the cheapest tonne.

Hong Kong – October 21, 2026 – (AsiaPressWire) – In May 2026, a Boston Consulting Group report carried a line that the city’s financial establishment had been working toward for three years: Hong Kong had overtaken Switzerland as the world’s largest cross-border wealth management centre. Cross-boundary wealth booked there rose 10.7 percent in 2025 to roughly USD 2.9 trillion, and the report projected about 9 percent annual growth through 2030. That is a real achievement, and it deserves scrutiny rather than celebration. This piece is about what the headline numbers actually measure, how the tax concession really works — including the parts brochures omit — what the 2026 legislative amendment would change, and where the genuine risks sit for a family deciding whether to set up there. Separating three numbers that get conflated Most coverage of this market mixes together three figures that measure entirely different things. Knowing which is which will save you from a bad decision. USD 2.9 trillion is cross-boundary wealth booked in Hong Kong. It is a BCG measure, and it is the number behind the “largest” claim. It captures wealth held by people who are not domiciled in the place where it is booked — which is why it can exceed the entire domestic wealth of many countries. HK$35.1 trillion (about USD 4.53 trillion) is industry-wide assets under management reported by the Securities and Futures Commission at the end of 2024. This is the whole asset and wealth management industry, not family office assets, and it should never be described as money managed by family offices. 3,384 single family offices were operating in Hong Kong at the end of 2025, according to a market study commissioned by Invest Hong Kong and conducted by Deloitte and published in February 2026. That is 681 more than at the end of 2023 — an increase of more than 25 percent over two years. The third figure is the one that matters for anyone actually setting up. The first two describe a financial centre; the third describes an operating ecosystem. The study also estimated that single family offices contribute about HK$12.6 billion a year through operating expenditure and directly employ more than 10,000 full-time professionals. Those are real-economy numbers — salaries, leases, audit and legal fees — rather than assets parked in a jurisdiction. How the tax concession actually works The concessionary profits tax rate for an eligible family-owned investment holding vehicle, or FIHV, is 0 percent on assessable profits from qualifying transactions and permitted incidental transactions. The regime applies to years of assessment beginning on or after 1 April 2022. The conditions are where applications either succeed or quietly fail. RequirementSpecific thresholdWhy it trips people upMinimum assetsHK$240 million aggregate net asset valueMust be specified assets managed for the relevant FIHV(s)StaffingAt least 2 qualified full-time employeesMust be adequate for the scale and complexity of activityOperating expenditureAt least HK$2 million in Hong KongAnnual, and adequacy is assessed against actual activityManagementManaged by an eligible single-family office in Hong KongThe office itself must qualifyOwnershipFamily holds at least 95% beneficial interestSubject to specific rules for charitable entities and unrelated personsSubstanceCore income-generating activities conducted in Hong KongNot a holding company on paper Hong Kong also applies a territorial tax system: capital gains are not taxed, offshore profits are not taxed when the applicable sourcing rules are met, and dividend income is generally not taxable. Each of those still requires classification and source analysis. “No tax” is a conclusion you reach after an analysis, not a starting assumption. The most important caution, and the one most often missed in marketing: the concession is not automatic. Families must satisfy ownership, asset, management, substance, and transaction conditions. Structures should be tested before assets are transferred into them — not after. The 2026 amendment bill, described carefully The government gazetted an amendment bill on 12 June 2026. It proposes to: Expand the definition of “fund” to cover retirement funds and donation funds Expand qualifying investments to include offshore real estate, carbon emission allowances, digital assets, precious metals, and private credit or loans Remove the 5 percent threshold for incidental transactions Relax parts of the special-purpose-entity treatment Enhance the carried-interest regime This matters directly for portfolios holding private credit, loans, digital assets, precious metals, and other non-traditional investments — the kinds of assets that younger principals increasingly hold and that the current rules treat awkwardly. One point that must be stated plainly: as of late August 2026 the bill had not been enacted. The Bills Committee had completed clause-by-clause examination, and the government aimed to resume the second reading debate in the second half of 2026. Families should model both the law currently in force and the proposed rules, and implement against the final enacted text and any Inland Revenue Department guidance — not against headlines, consultation summaries, or a service provider’s slide deck. What actually attracts families here Tax is the headline, but it is rarely the deciding factor once a family has been advised properly. Four things do the real work. Access to Mainland China and the Greater Bay Area. Stock Connect links the Hong Kong and Mainland securities markets. The Cross-boundary Wealth Management Connect scheme allows eligible residents in Hong Kong, Macao, and Mainland Greater Bay Area cities to invest in approved wealth-management products across the boundary. These programmes do not give every family office unrestricted access to every Mainland asset — eligibility, product scope, quotas, account arrangements, and regulatory requirements all still apply — but their strategic value is the presence of regulated channels inside a broader cross-border financial ecosystem. The offshore renminbi pool. Hong Kong holds the deepest offshore renminbi liquidity pool outside Mainland China, according to the Hong Kong Monetary Authority. For families with renminbi assets, Mainland operating businesses, trade flows, or anticipated future investment needs, this is a practical treasury advantage rather than a talking point. Common law and succession. Hong Kong’s legal system is based on common law with judicial independence, distinct from the legal system of Mainland China under the “one country, two systems” framework. For families thinking about ownership structures spanning generations and jurisdictions, the predictability of the forum matters more than the tax rate. The ecosystem. The government met, ahead of schedule in September 2025, a target set in the 2022 Policy Address to help at least 200 family offices establish or expand in the city, and has set a further target of at least 220 more between 2026 and 2028. Supporting infrastructure includes the Hong Kong Academy for Wealth Legacy, which has run over 20 events attracting around 3,100 family asset owners and next-generation leaders; the Impact Link platform for directing family capital toward social outcomes; and international linkages including a “Hong Kong Family Office Nexus” with Bloomberg and an agreement with a UAE family office association extending the network into the Middle East. Where families get this wrong Treating the tax concession as automatic. It is not, and discovering the substance requirements after incorporation is expensive. Assuming “no capital gains tax” ends the analysis. Source analysis is still required for every income stream. Classification errors surface years later, during a review. Setting up the structure before testing it. Transferring assets into a structure that later fails a condition is far more costly than modelling it first. Confusing AUM with family office assets. Presenting the industry’s HK$35.1 trillion as family office money is a category error that misleads nobody who knows the market. Overlooking succession as the real driver. Families that set up for tax reasons alone tend to build structures that do not survive the generational transfer — which is the thing the structure was supposedly for. What the concession does not do The concessionary rate is narrower than its reputation, and the gap between the two is where most disappointment originates. Six limits are worth stating explicitly. It applies to the vehicle, not to the family. Profits tax on the FIHV’s qualifying transactions is what is exempted. It does not change the personal tax residence of any family member, and it does nothing about the tax position of an operating business held outside the vehicle. Families sometimes assume that establishing an FIHV shelters trading profits from a factory or a retail chain; it does not, and attempting to route operating revenue through the structure is a fast way to fail the qualifying transaction test. It does not eliminate reporting. Common Reporting Standard and FATCA obligations are driven by residence and account classification, not by whether a vehicle enjoys a concessionary rate. Beneficial ownership registers, economic substance filings, and home-jurisdiction disclosure rules continue to apply. A structure that reduces Hong Kong profits tax to zero can still be fully visible to a family’s home tax authority, and should be planned on that assumption. It does not override succession law elsewhere. Hong Kong’s common-law forum governs what it governs. It does not displace forced heirship rules, community property regimes, or matrimonial provisions in the jurisdictions where family members are domiciled or hold assets. Cross-border succession planning still has to be done jurisdiction by jurisdiction. It does not exempt salaries tax. The qualified employees working in the office pay salaries tax like anyone else, and the employer carries the usual obligations. It does not remove stamp duty. Transfers of Hong Kong real property and Hong Kong shares attract stamp duty under the general rules; families moving local property into a structure should price that before committing. Finally, it does not remove the need for advice. Every condition described above is tested against facts, and facts differ by family. Questions families actually ask Do I need to incorporate a local entity? Typically yes, since the FIHV must be managed by an eligible single-family office in Hong Kong and must conduct core income-generating activities there. Can charity or non-family holdings be included? The general rule requires the family to hold at least 95 percent beneficial interest, subject to specific rules for charitable entities and unrelated persons. Specialised advice is required rather than a general assumption. How long does setup take? The licensing and structuring sequence typically runs several months, and the two-employee plus HK$2 million expenditure requirements must be genuinely met rather than arranged on paper. Should we wait for the amendment bill to pass? No. Model both regimes, and implement against enacted law. Waiting has a cost too, and the bill’s direction — broader qualifying assets — is unlikely to make current compliant structures worse. What Hong Kong is actually competing on Calling this a contest between Hong Kong and Switzerland or Singapore is a simplification that hides more than it reveals. Different centres win on different mandates. Switzerland retains strengths in private banking depth, political stability, and a long-established reputation for discretion — the reasons it held the top cross-border position for decades before this ranking. Competition remains real, and the ranking may move again. The relevant question for a family is narrower: does this jurisdiction give regulated access to the markets they actually transact in, a legal system that handles their succession plan predictably, and a talent pool to staff the office? For families whose operating businesses, investments, or family members are concentrated in Asia, particularly with Mainland exposure, Hong Kong answers those three questions unusually well. For families with no Asian exposure, the answer may be different, and pretending otherwise serves no one. What setup actually costs, and how long it takes Brochures quote the tax rate and stop. A family planning properly needs the operating budget, because the concession has to be maintained year after year rather than granted once. Cost lineIndicative rangeNotesMinimum qualifying assetsHK$240 millionNet asset value of specified assetsMinimum annual operating expenditureHK$2 millionMust be genuine Hong Kong spend, assessed for adequacyMinimum staffing2 qualified full-time employeesMust be adequate to scale and complexityStructuring and legalLow-to-mid six figures HKDDepends on asset classes and jurisdictions involvedAnnual audit, compliance, administrationMid six figures HKD recurringScales with portfolio complexityOffice premisesMarket rateSubstance expectations make a genuine office necessary The recurring spend is the part families underestimate. Two qualified full-time employees plus HK$2 million of operating expenditure plus audit and administration means a real annual budget in the low millions of Hong Kong dollars. For a family with HK$240 million in specified assets, that is roughly one percent of assets per year before any investment cost — reasonable relative to most wealth management pricing, but not trivial, and it is the reason the concession is not worth pursuing for smaller portfolios. On timing, the sequence typically runs several months: structure design and asset mapping first, then incorporation and office setup, then staffing, then application and substantiation. The most common cause of delay is discovering late that assets intended for the vehicle are not “specified assets” under the rules, or that the family’s ownership structure does not satisfy the 95 percent test without reorganisation. Staffing the office: the constraint nobody budgets for Two qualified full-time employees is the statutory floor, and treating it as a staffing plan is one of the most common planning errors. The requirement is not simply two people on a payroll; it is two people whose qualifications are adequate for the scale and complexity of the activity actually being carried on. A vehicle holding HK$240 million of listed equities and a vehicle holding private credit, offshore property, and digital assets are not the same undertaking, and the same headcount will not look adequate for both. In practice, most offices that want genuine in-house capability land between three and five people. One is an investment professional with a track record relevant to the asset classes held. One covers compliance, accounting, and the substantiation file that the Inland Revenue Department may ask to see. A third handles operations — custody relationships, reporting, and the administrative load of a multi-jurisdiction portfolio. Below that, the office becomes a mailbox, and the substance test starts to look fragile. Cost is the obvious issue, and availability is the less obvious one. Adding 220 more family offices to the city by 2028 competes for the same pool of private wealth professionals that banks, asset managers, and trustees are already drawing from. Families arriving now are hiring into a reasonably tight market; families arriving in 2028 will likely find it tighter. Budget for recruitment time as well as salary, and expect the search for the compliance hire to take longer than the search for the investment hire. One further point on outsourcing. Using external asset managers is entirely normal and does not by itself break the substance requirement. But if substantially all investment decisions are made outside Hong Kong, the assertion that core income-generating activities are conducted in Hong Kong becomes difficult to sustain. The defensible arrangement is decision-making authority resident in the office, with external managers executing within mandates the office sets and monitors. That is a governance design question, and it should be settled before the application rather than after a query arrives. Three archetypes, three different answers Whether Hong Kong is the right choice depends heavily on what kind of family is asking. The Asia-operating family. A family whose operating business is in Mainland China or Southeast Asia, with renminbi and US dollar flows, next-generation members likely to study or work in the region, and a succession event five to fifteen years out. For this family, the combination of regulated Mainland access, offshore renminbi depth, and a common-law succession forum is genuinely hard to replicate. Hong Kong is usually the right answer. The globally diversified family. A family with assets across Europe, North America, and Asia, no single jurisdictional concentration, and an existing structure elsewhere that works. For this family, adding a Hong Kong FIHV is a diversification of jurisdiction rather than a core move, and it only makes sense if there is a real reason — Mainland exposure, an Asian operating business, or family members relocating. Adding a structure because the tax rate is zero is how families end up maintaining entities they do not need. The mobile principal. An entrepreneur who has recently exited, holds a meaningful share of the proceeds in digital assets or private credit, and may relocate more than once over the next decade. This family benefits most from the 2026 amendment if it passes, because the current qualifying asset list sits awkwardly with newer asset classes. It is also the family most exposed to the risk of planning against proposed rather than enacted rules — the profile that most needs advice grounded in current law. Glossary FIHV — Family-owned investment holding vehicle, the entity eligible for the concessionary rate. Specified assets — The asset classes that count toward both the HK$240 million threshold and the qualifying transaction test. Incidental transactions — Non-qualifying transactions permitted up to a threshold; the 2026 bill proposes removing the 5 percent limit. Substance — The requirement that core income-generating activities genuinely occur in Hong Kong, not merely that an entity is registered there. Source analysis — The determination of where profits arise, which decides whether the territorial system exempts them. Forward calls, and the risks nobody prices in The five-year plan’s stated vision is to comprehensively surpass Switzerland as the world’s largest wealth management centre by the end of 2030, with the government explicitly framing the next phase as “family office 2.0” — a shift from expanding the count of offices to deepening the ecosystem. That framing is more credible than a raw numbers race, because ecosystem depth is what survives a change in tax rules elsewhere. Three risks deserve attention. First, the ranking depends partly on capital flows and market performance that no policy controls; a weak year for Mainland listings moves the number. Second, the 2026 amendment’s passage is not guaranteed in its current form, and planning against proposed rather than enacted rules is a recurring source of expensive rework. Third, geopolitics cuts both ways: it has reinforced Hong Kong’s role as a safe harbour within the Chinese financial system for some families while raising questions for others whose own home jurisdictions view the exposure differently. One further risk is subtle and worth naming: capacity. Adding 220 more family offices by 2028 requires experienced private wealth professionals, auditors, and trustees. If the talent pipeline lags the inflow, service quality and fees will both adjust, and the families that arrive last will feel it most. A checklist before you commit Test the ownership, asset, staffing, and expenditure conditions against your actual structure — before transferring assets Confirm the two employees are genuinely qualified and the HK$2 million spend is real annual operating expenditure Source-analyse every income stream; do not assume “offshore means exempt” Model both the current law and the proposed 2026 amendments Decide whether your real driver is tax, succession, or market access — and check the structure serves that driver Ask your adviser which jurisdiction they would recommend if your assets were entirely outside Asia, and why The honest bottom line The 3,384 figure and the top ranking are genuine. They describe a financial centre that has deliberately built an ecosystem — tax concessions, a new capital investment entrant scheme, an academy, philanthropic infrastructure, and international agreements — and has done so fast enough to move a global ranking within three years. But a ranking is a snapshot, and a tax rate is a starting condition rather than an outcome. The families who do well here are not the ones who moved for the 0 percent headline. They are the ones who tested their structure against the substance rules first, understood that succession and market access were the actual reasons to be there, and accepted that the concession is something you qualify for continuously — not something you are granted once.

Logan Pierce – AsiaPresswire – In May 2026, a Boston Consulting Group report carried a line that the city’s financial establishment had been working toward for three years: Hong Kong had overtaken Switzerland as the world’s largest cross-border wealth management centre. Cross-boundary wealth booked there rose 10.7 percent in 2025 to roughly USD 2.9 trillion, and the report projected about 9 percent annual growth through 2030. That is a real achievement, and it deserves scrutiny rather than celebration. This piece is about what the headline numbers actually measure, how the tax concession really works — including the parts brochures omit — what the 2026 legislative amendment would change, and where the genuine risks sit for a family deciding whether to set up there. Separating three numbers that get conflated Most coverage of this market mixes together three figures that measure entirely different things. Knowing which is which will save you from a bad decision. USD 2.9 trillion is cross-boundary wealth booked in Hong Kong. It is a BCG measure, and it is the number behind the “largest” claim. It captures wealth held by people who are not domiciled in the place where it is booked — which is why it can exceed the entire domestic wealth of many countries. HK$35.1 trillion (about USD 4.53 trillion) is industry-wide assets under management reported by the Securities and Futures Commission at the end of 2024. This is the whole asset and wealth management industry, not family office assets, and it should never be described as money managed by family offices. 3,384 single family offices were operating in Hong Kong at the end of 2025, according to a market study commissioned by Invest Hong Kong and conducted by Deloitte and published in February 2026. That is 681 more than at the end of 2023 — an increase of more than 25 percent over two years. The third figure is the one that matters for anyone actually setting up. The first two describe a financial centre; the third describes an operating ecosystem. The study also estimated that single family offices contribute about HK$12.6 billion a year through operating expenditure and directly employ more than 10,000 full-time professionals. Those are real-economy numbers — salaries, leases, audit and legal fees — rather than assets parked in a jurisdiction. How the tax concession actually works The concessionary profits tax rate for an eligible family-owned investment holding vehicle, or FIHV, is 0 percent on assessable profits from qualifying transactions and permitted incidental transactions. The regime applies to years of assessment beginning on or after 1 April 2022. The conditions are where applications either succeed or quietly fail. Requirement Specific threshold Why it trips people up Minimum assets HK$240 million aggregate net asset value Must be specified assets managed for the relevant FIHV(s) Staffing At least 2 qualified full-time employees Must be adequate for the scale and complexity of activity Operating expenditure At least HK$2 million in Hong Kong Annual, and adequacy is assessed against actual activity Management Managed by an eligible single-family office in Hong Kong The office itself must qualify Ownership Family holds at least 95% beneficial interest Subject to specific rules for charitable entities and unrelated persons Substance Core income-generating activities conducted in Hong Kong Not a holding company on paper Hong Kong also applies a territorial tax system: capital gains are not taxed, offshore profits are not taxed when the applicable sourcing rules are met, and dividend income is generally not taxable. Each of those still requires classification and source analysis. “No tax” is a conclusion you reach after an analysis, not a starting assumption. The most important caution, and the one most often missed in marketing: the concession is not automatic. Families must satisfy ownership, asset, management, substance, and transaction conditions. Structures should be tested before assets are transferred into them — not after. The 2026 amendment bill, described carefully The government gazetted an amendment bill on 12 June 2026. It proposes to: Expand the definition of “fund” to cover retirement funds and donation funds Expand qualifying investments to include offshore real estate, carbon emission allowances, digital assets, precious metals, and private credit or loans Remove the 5 percent threshold for incidental transactions Relax parts of the special-purpose-entity treatment Enhance the carried-interest regime This matters directly for portfolios holding private credit, loans, digital assets, precious metals, and other non-traditional investments — the kinds of assets that younger principals increasingly hold and that the current rules treat awkwardly. One point that must be stated plainly: as of late August 2026 the bill had not been enacted. The Bills Committee had completed clause-by-clause examination, and the government aimed to resume the second reading debate in the second half of 2026. Families should model both the law currently in force and the proposed rules, and implement against the final enacted text and any Inland Revenue Department guidance — not against headlines, consultation summaries, or a service provider’s slide deck. What actually attracts families here Tax is the headline, but it is rarely the deciding factor once a family has been advised properly. Four things do the real work. Access to Mainland China and the Greater Bay Area. Stock Connect links the Hong Kong and Mainland securities markets. The Cross-boundary Wealth Management Connect scheme allows eligible residents in Hong Kong, Macao, and Mainland Greater Bay Area cities to invest in approved wealth-management products across the boundary. These programmes do not give every family office unrestricted access to every Mainland asset — eligibility, product scope, quotas, account arrangements, and regulatory requirements all still apply — but their strategic value is the presence of regulated channels inside a broader cross-border financial ecosystem. The offshore renminbi pool. Hong Kong holds the deepest offshore renminbi liquidity pool outside Mainland China, according to the Hong Kong Monetary Authority. For families with renminbi assets, Mainland operating businesses, trade flows, or anticipated future investment needs, this is a practical treasury advantage rather than a talking point. Common law and succession. Hong Kong’s legal system is based on common law with judicial independence, distinct from the legal system of Mainland China under the “one country, two systems” framework. For families thinking about ownership structures spanning generations and jurisdictions, the predictability of the forum matters more than the tax rate. The ecosystem. The government met, ahead of schedule in September 2025, a target set in the 2022 Policy Address to help at least 200 family offices establish or expand in the city, and has set a further target of at least 220 more between 2026 and 2028. Supporting infrastructure includes the Hong Kong Academy for Wealth Legacy, which has run over 20 events attracting around 3,100 family asset owners and next-generation leaders; the Impact Link platform for directing family capital toward social outcomes; and international linkages including a “Hong Kong Family Office Nexus” with Bloomberg and an agreement with a UAE family office association extending the network into the Middle East. Where families get this wrong Treating the tax concession as automatic. It is not, and discovering the substance requirements after incorporation is expensive. Assuming “no capital gains tax” ends the analysis. Source analysis is still required for every income stream. Classification errors surface years later, during a review. Setting up the structure before testing it. Transferring assets into a structure that later fails a condition is far more costly than modelling it first. Confusing AUM with family office assets. Presenting the industry’s HK$35.1 trillion as family office money is a category error that misleads nobody who knows the market. Overlooking succession as the real driver. Families that set up for tax reasons alone tend to build structures that do not survive the generational transfer — which is the thing the structure was supposedly for. What the concession does not do The concessionary rate is narrower than its reputation, and the gap between the two is where most disappointment originates. Six limits are worth stating explicitly. It applies to the vehicle, not to the family. Profits tax on the FIHV’s qualifying transactions is what is exempted. It does not change the personal tax residence of any family member, and it does nothing about the tax position of an operating business held outside the vehicle. Families sometimes assume that establishing an FIHV shelters trading profits from a factory or a retail chain; it does not, and attempting to route operating revenue through the structure is a fast way to fail the qualifying transaction test. It does not eliminate reporting. Common Reporting Standard and FATCA obligations are driven by residence and account classification, not by whether a vehicle enjoys a concessionary rate. Beneficial ownership registers, economic substance filings, and home-jurisdiction disclosure rules continue to apply. A structure that reduces Hong Kong profits tax to zero can still be fully visible to a family’s home tax authority, and should be planned on that assumption. It does not override succession law elsewhere. Hong Kong’s common-law forum governs what it governs. It does not displace forced heirship rules, community property regimes, or matrimonial provisions in the jurisdictions where family members are domiciled or hold assets. Cross-border succession planning still has to be done jurisdiction by jurisdiction. It does not exempt salaries tax. The qualified employees working in the office pay salaries tax like anyone else, and the employer carries the usual obligations. It does not remove stamp duty. Transfers of Hong Kong real property and Hong Kong shares attract stamp duty under the general rules; families moving local property into a structure should price that before committing. Finally, it does not remove the need for advice. Every condition described above is tested against facts, and facts differ by family. Questions families actually ask Do I need to incorporate a local entity? Typically yes, since the FIHV must be managed by an eligible single-family office in Hong Kong and must conduct core income-generating activities there. Can charity or non-family holdings be included? The general rule requires the family to hold at least 95 percent beneficial interest, subject to specific rules for charitable entities and unrelated persons. Specialised advice is required rather than a general assumption. How long does setup take? The licensing and structuring sequence typically runs several months, and the two-employee plus HK$2 million expenditure requirements must be genuinely met rather than arranged on paper. Should we wait for the amendment bill to pass? No. Model both regimes, and implement against enacted law. Waiting has a cost too, and the bill’s direction — broader qualifying assets — is unlikely to make current compliant structures worse. What Hong Kong is actually competing on Calling this a contest between Hong Kong and Switzerland or Singapore is a simplification that hides more than it reveals. Different centres win on different mandates. Switzerland retains strengths in private banking depth, political stability, and a long-established reputation for discretion — the reasons it held the top cross-border position for decades before this ranking. Competition remains real, and the ranking may move again. The relevant question for a family is narrower: does this jurisdiction give regulated access to the markets they actually transact in, a legal system that handles their succession plan predictably, and a talent pool to staff the office? For families whose operating businesses, investments, or family members are concentrated in Asia, particularly with Mainland exposure, Hong Kong answers those three questions unusually well. For families with no Asian exposure, the answer may be different, and pretending otherwise serves no one. What setup actually costs, and how long it takes Brochures quote the tax rate and stop. A family planning properly needs the operating budget, because the concession has to be maintained year after year rather than granted once. Cost line Indicative range Notes Minimum qualifying assets HK$240 million Net asset value of specified assets Minimum annual operating expenditure HK$2 million Must be genuine Hong Kong spend, assessed for adequacy Minimum staffing 2 qualified full-time employees Must be adequate to scale and complexity Structuring and legal Low-to-mid six figures HKD Depends on asset classes and jurisdictions involved Annual audit, compliance, administration Mid six figures HKD recurring Scales with portfolio complexity Office premises Market rate Substance expectations make a genuine office necessary The recurring spend is the part families underestimate. Two qualified full-time employees plus HK$2 million of operating expenditure plus audit and administration means a real annual budget in the low millions of Hong Kong dollars. For a family with HK$240 million in specified assets, that is roughly one percent of assets per year before any investment cost — reasonable relative to most wealth management pricing, but not trivial, and it is the reason the concession is not worth pursuing for smaller portfolios. On timing, the sequence typically runs several months: structure design and asset mapping first, then incorporation and office setup, then staffing, then application and substantiation. The most common cause of delay is discovering late that assets intended for the vehicle are not “specified assets” under the rules, or that the family’s ownership structure does not satisfy the 95 percent test without reorganisation. Staffing the office: the constraint nobody budgets for Two qualified full-time employees is the statutory floor, and treating it as a staffing plan is one of the most common planning errors. The requirement is not simply two people on a payroll; it is two people whose qualifications are adequate for the scale and complexity of the activity actually being carried on. A vehicle holding HK$240 million of listed equities and a vehicle holding private credit, offshore property, and digital assets are not the same undertaking, and the same headcount will not look adequate for both. In practice, most offices that want genuine in-house capability land between three and five people. One is an investment professional with a track record relevant to the asset classes held. One covers compliance, accounting, and the substantiation file that the Inland Revenue Department may ask to see. A third handles operations — custody relationships, reporting, and the administrative load of a multi-jurisdiction portfolio. Below that, the office becomes a mailbox, and the substance test starts to look fragile. Cost is the obvious issue, and availability is the less obvious one. Adding 220 more family offices to the city by 2028 competes for the same pool of private wealth professionals that banks, asset managers, and trustees are already drawing from. Families arriving now are hiring into a reasonably tight market; families arriving in 2028 will likely find it tighter. Budget for recruitment time as well as salary, and expect the search for the compliance hire to take longer than the search for the investment hire. One further point on outsourcing. Using external asset managers is entirely normal and does not by itself break the substance requirement. But if substantially all investment decisions are made outside Hong Kong, the assertion that core income-generating activities are conducted in Hong Kong becomes difficult to sustain. The defensible arrangement is decision-making authority resident in the office, with external managers executing within mandates the office sets and monitors. That is a governance design question, and it should be settled before the application rather than after a query arrives. Three archetypes, three different answers Whether Hong Kong is the right choice depends heavily on what kind of family is asking. The Asia-operating family. A family whose operating business is in Mainland China or Southeast Asia, with renminbi and US dollar flows, next-generation members likely to study or work in the region, and a succession event five to fifteen years out. For this family, the combination of regulated Mainland access, offshore renminbi depth, and a common-law succession forum is genuinely hard to replicate. Hong Kong is usually the right answer. The globally diversified family. A family with assets across Europe, North America, and Asia, no single jurisdictional concentration, and an existing structure elsewhere that works. For this family, adding a Hong Kong FIHV is a diversification of jurisdiction rather than a core move, and it only makes sense if there is a real reason — Mainland exposure, an Asian operating business, or family members relocating. Adding a structure because the tax rate is zero is how families end up maintaining entities they do not need. The mobile principal. An entrepreneur who has recently exited, holds a meaningful share of the proceeds in digital assets or private credit, and may relocate more than once over the next decade. This family benefits most from the 2026 amendment if it passes, because the current qualifying asset list sits awkwardly with newer asset classes. It is also the family most exposed to the risk of planning against proposed rather than enacted rules — the profile that most needs advice grounded in current law. Glossary FIHV — Family-owned investment holding vehicle, the entity eligible for the concessionary rate. Specified assets — The asset classes that count toward both the HK$240 million threshold and the qualifying transaction test. Incidental transactions — Non-qualifying transactions permitted up to a threshold; the 2026 bill proposes removing the 5 percent limit. Substance — The requirement that core income-generating activities genuinely occur in Hong Kong, not merely that an entity is registered there. Source analysis — The determination of where profits arise, which decides whether the territorial system exempts them. Forward calls, and the risks nobody prices in The five-year plan’s stated vision is to comprehensively surpass Switzerland as the world’s largest wealth management centre by the end of 2030, with the government explicitly framing the next phase as “family office 2.0” — a shift from expanding the count of offices to deepening the ecosystem. That framing is more credible than a raw numbers race, because ecosystem depth is what survives a change in tax rules elsewhere. Three risks deserve attention. First, the ranking depends partly on capital flows and market performance that no policy controls; a weak year for Mainland listings moves the number. Second, the 2026 amendment’s passage is not guaranteed in its current form, and planning against proposed rather than enacted rules is a recurring source of expensive rework. Third, geopolitics cuts both ways: it has reinforced Hong Kong’s role as a safe harbour within the Chinese financial system for some families while raising questions for others whose own home jurisdictions view the exposure differently. One further risk is subtle and worth naming: capacity. Adding 220 more family offices by 2028 requires experienced private wealth professionals, auditors, and trustees. If the talent pipeline lags the inflow, service quality and fees will both adjust, and the families that arrive last will feel it most. A checklist before you commit Test the ownership, asset, staffing, and expenditure conditions against your actual structure — before transferring assets Confirm the two employees are genuinely qualified and the HK$2 million spend is real annual operating expenditure Source-analyse every income stream; do not assume “offshore means exempt” Model both the current law and the proposed 2026 amendments Decide whether your real driver is tax, succession, or market access — and check the structure serves that driver Ask your adviser which jurisdiction they would recommend if your assets were entirely outside Asia, and why The honest bottom line The 3,384 figure and the top ranking are genuine. They describe a financial centre that has deliberately built an ecosystem — tax concessions, a new capital investment entrant scheme, an academy, philanthropic infrastructure, and international agreements — and has done so fast enough to move a global ranking within three years. But a ranking is a snapshot, and a tax rate is a starting condition rather than an outcome. The families who do well here are not the ones who moved for the 0 percent headline. They are the ones who tested their structure against the substance rules first, understood that succession and market access were the actual reasons to be there, and accepted that the concession is something you qualify for continuously — not something you are granted once.

Robert Sterling – AsiaPresswire – Indonesia banned the export of unprocessed nickel ore in January 2020, pulling forward a deadline that had originally been set for 2022. Five years on, it is the most consequential industrial policy experiment in Southeast Asia, and the clearest test anywhere of whether a commodity-producing country can force its way up the value chain by legislating against its own exports. The answer is genuinely mixed, and the mix is more instructive than either the celebratory or the dismissive version of the story. Indonesia did build a processing industry that did not exist. It also created chronic global oversupply, locked itself into a technology and capital structure it does not control, and discovered that banning an export is much easier than capturing the margin that sits at the end of the chain. This piece is about what the policy actually built, what it cost, where the money actually goes, and what a company deciding whether to commit capital there needs to test before it writes a cheque. The policy, stated precisely The measure most people mean by “downstreaming” is a prohibition on exporting nickel ore, introduced through a mineral and coal mining regulation and brought forward to take effect in January 2020. The logic was straightforward: if ore cannot leave, processors must come to the ore. Three things about it are worth being precise about, because the shorthand versions mislead. It is not a tax. It is a prohibition. That distinction matters for modelling, because a tax can be priced and hedged while a prohibition reorders the physical geography of an industry. It is not limited to nickel. Indonesia has applied the same logic to bauxite and has extended the approach, in various forms, to other commodities including copper and tin. Companies treating this as a nickel-specific rule are reading the policy too narrowly. It is not static. Thresholds, permitted product forms, royalty structures, and domestic-market obligations have been adjusted repeatedly since 2020. Any model built on the rules as they stood in a given year needs re-checking against the rules as they stand now. The instrument has also been paired with inducements: tax holidays, investment allowances, and the designation of industrial estates where processing is concentrated. The two largest are the Morowali industrial park on Sulawesi and the Weda Bay complex on Halmahera in North Maluku. Most of the country’s processing capacity sits inside those two sites, which is itself a concentration risk worth naming. What the country actually became On the narrow question of whether the policy worked as industrial policy — did processing capacity appear? — the answer is unambiguously yes. Indonesia moved from exporting raw ore to exporting ferronickel, nickel pig iron, matte, mixed hydroxide precipitate, and, further along, nickel sulfate and precursor material. By the middle of the decade it accounted for a share of global mined nickel supply approaching or exceeding half, depending on the year and the source consulted, up from a fraction of that before the ban. Export revenue from nickel products rose from low single-digit billions of US dollars in the ore era to tens of billions afterwards — a genuine, large, and fast shift in the composition of the country’s exports. Employment at the two main parks runs to tens of thousands of direct jobs. That is a real number in a country where the alternative use of remote land in Sulawesi and Halmahera was close to nil. What is less often stated precisely is what that capacity is for. The overwhelming majority of Indonesian processing capacity was built for stainless steel — ferronickel and nickel pig iron feeding the stainless chain — rather than for batteries. The battery-grade portion, produced mostly through high-pressure acid leaching into mixed hydroxide precipitate and then into sulfate, is real and growing but remains the smaller part of the complex. A country that banned ore exports to serve the electric vehicle revolution ended up, in its first phase, mostly serving stainless steel. Where the chain actually stops Stage What it produces Where Indonesia sits What it captures Mining Laterite and saprolite ore Dominant global position Lowest margin, highest physical footprint Smelting (RKEF) Ferronickel, nickel pig iron Very large, largely built Commodity margin, tied to stainless demand Refining (HPAL) Mixed hydroxide precipitate Growing, capital-intensive Higher margin, exposed to cobalt and acid costs Sulfate and precursor Battery-grade nickel chemicals Present but limited Approaching cell-grade pricing Cathode and cell Cathode active material, cells Early / under construction The margin everyone wants Pack and vehicle Battery packs, EVs Assembly only The margin downstreaming was aimed at The table is the argument. Indonesia is unambiguously dominant at row one, very strong at row two, present at row three, limited at row four, and not yet established at rows five and six. The stated ambition of the policy was row six. The delivered result, five years in, is mostly rows one to three. This is not a criticism of execution so much as a description of what happened. Each step up the table requires a different input — not more ore, but technology, process know-how, qualification with cell makers, and access to a customer who will accept your material into a battery supply chain. Those are not things a prohibition on ore exports can compel. They have to be bought, licensed, or built over years. The stainless anchor, and why it still sets the floor Battery demand gets the attention, but stainless steel still absorbs the majority of primary nickel, and that fact is what makes the Indonesian build-out financially legible at all. RKEF lines producing ferronickel and nickel pig iron have a customer base that exists today, in volume, and does not require qualification. That is why the first phase of downstreaming went there rather than into battery chemicals: the capital was available, the technology was mature, the off-taker was known, and the product could be sold the day the line started. This has a stabilising effect that is easy to miss. When battery-grade nickel prices weaken or when a cell maker delays a programme, the stainless chain keeps absorbing metal. It acts as a floor under offtake, and therefore under utilisation, in a way that a purely battery-facing industry would not have. It also means that the economics of most Indonesian capacity are driven first by stainless demand in China and India and only second by electric vehicle forecasts — which is the opposite of the framing most coverage uses. The consequence for anyone modelling this market is that “nickel demand” is not one number. It is two largely separate demand pools with different price sensitivities, different customers, and different contract structures, connected mainly by the fact that producers can switch some capacity between them. Treating them as a single market is why so many forecasts of the 2021–2023 period were wrong in the same direction. What it cost The costs are real and they fall in four places. Capital. Smelting and refining are among the most capital-intensive industrial activities in the commodity world. A rotary kiln–electric furnace line is expensive; a high-pressure acid leach plant is more expensive still, and its economics depend on ore grade, acid supply, and the cobalt credit in the byproduct stream. The billions of dollars invested came overwhelmingly from foreign capital, and the single largest source of that capital is Chinese. Power. Indonesian smelting is heavily coal-fired, with captive power plants built alongside the lines. That has two consequences. The carbon intensity of the resulting nickel is high, which increasingly matters for buyers who have to report it. And it ties the economics of the metal to the cost and availability of domestic coal, which the government also prices and regulates. Environment. Laterite mining clears land and moves very large volumes of material. The tailings question is the hardest one, and in parts of the country the practice of disposing of tailings at sea has drawn sustained criticism. Acid leaching creates its own waste stream. Companies sourcing Indonesian nickel and reporting against European battery or supply-chain due diligence rules should assume these issues will be raised, and should have sourcing documentation that withstands it. Market structure. This is the cost that was least anticipated. Indonesia added processing capacity faster than global nickel demand grew. The result, visible through 2023 and into the following years, was sustained oversupply and a nickel price that fell well below the level at which the newest and highest-cost capacity is comfortable. Parts of the industry outside Indonesia — Australian laterite operations and several projects elsewhere — were squeezed or shut. Then the pressure came home: margins inside Indonesia compressed as well. The part the brochures skip: who captures the margin A prohibition on ore exports guarantees that processing happens in the country. It does not guarantee that the profit from processing accrues to the country or to its companies. The ownership structure of much of the built capacity is foreign. Offtake is frequently tied to the same groups that provided the capital. Financing, engineering procurement, and construction were largely supplied from outside. In that arrangement, Indonesia receives employment, taxes and royalties, the depreciation of a large physical asset base, and the strategic position of being indispensable to a critical supply chain. Those are not nothing — employment and fiscal receipts are the point of industrial policy — but they are not the same as capturing the margin at the cell and vehicle end of the table. There is a second-order effect here that deserves attention. Because the dominant capital and offtake relationships run through Chinese groups, Indonesian nickel has become entangled in the classification rules of other countries’ industrial policy. The United States clean-vehicle credit regime applies foreign-entity-of-concern restrictions that trace ownership and control through the supply chain. European rules apply different but overlapping tests. The practical consequence is that material which is perfectly saleable into one market may be disqualified from another, and the disqualification can be triggered by an ownership stake several levels up rather than by anything about the metal itself. For a company building a supply chain, this means the question is never simply “is there nickel available in Indonesia.” It is “which corporate entity produced it, who controls that entity, and where does the resulting material qualify.” Qualification: the step nobody budgets for Selling metal is one thing. Selling metal into a battery supply chain is a different and much slower business, and the gap between the two is where a great deal of announced capacity quietly stalls. Battery makers do not buy material on specification sheets. They qualify it, plant by plant, through a process that runs from months to well over a year. Samples are produced, cell-level testing is run against cycle life and safety behaviour, the plant’s process control is audited, and the material is then approved against a specific production line. Change the plant, change the process, or in some cases change the ore blend feeding it, and the qualification has to be revisited. This has three practical implications. First, a refinery that has not been qualified has no battery customer regardless of how good its product is on paper; it can only sell at commodity pricing, which is precisely the margin it was built to escape. Second, the bottleneck is not always chemistry — it is often simply the queue at the cell makers, who can only run a limited number of qualification programmes at once. Third, the value of an existing qualification is high and underappreciated in deal valuations, because it cannot be bought quickly at any price. There is a related trap on the buying side. A purchaser that signs an offtake for “battery-grade nickel” from a plant under construction is not buying battery-grade nickel; it is buying a schedule, a qualification process, and a set of risks that sit between the two. Contracts that do not distinguish those states — and that do not allocate the cost of a failed or delayed qualification — are the ones that end up in dispute. What other countries tried, and what happened Indonesia is not the first country to try this, and the comparison is instructive because the results differ in ways that track the specifics rather than the ambition. Chile has moved on lithium through a different instrument — a state-participation model for its salt flat operations, with the state taking control or majority stakes in new projects rather than a blanket export prohibition. The approach preserves more room for existing operators and leaves the question of where processing happens partly open. It also moves more slowly. Guinea holds the world’s largest bauxite reserves and has pushed for domestic alumina refining for years, with results that have been slower and more uneven than Indonesia’s, largely because the capital and infrastructure requirements are less forgiving and the bauxite market is less concentrated. Zimbabwe and Namibia both moved to restrict raw lithium exports, with Zimbabwe banning unprocessed lithium ore exports. The stated logic is identical to Indonesia’s. The scale of the resulting domestic industry, so far, is not — which is a reminder that a prohibition only works as industrial policy when there is capital ready to build against it and a customer willing to buy the output. The Philippines is the closest analogue and the most interesting case. It is a large nickel ore producer and has debated an export ban for years without implementing one, which has left it exporting ore into the Indonesian complex it might otherwise have competed with. The comparison suggests that the speed and decisiveness of Indonesia’s move mattered at least as much as the move itself. Across all of these, the pattern is consistent. Prohibitions work when three conditions hold: the country’s share of supply is large enough that the ban changes global trade rather than just diverting it, capital is available and willing to build on the country’s terms, and there is a market for the processed product that does not require the producer to win a qualification race first. Indonesia had all three for stainless-grade material and only the first two for battery-grade. Questions companies actually ask Can we get battery-grade material, or only stainless-grade? Both exist, but battery-grade capacity is the smaller share and is concentrated in specific plants. Qualification with a cell maker is a separate and slower process from buying metal. Is the ore export ban at risk of being reversed? Treat it as permanent policy direction, not a temporary measure. It has survived changes of administration and legal challenge. Model it as a structural feature of the market. What happens if the price stays low? Capacity built for a higher price does not disappear; it runs at lower margin, and consolidation follows. For a buyer, low prices are favourable. For an investor, they change the project’s whole return profile, and the projects most exposed are the newest and highest-cost ones. How do we handle the carbon intensity? You report it. The footprint of coal-fired smelting is materially higher than alternative supply, and buyers subject to reporting obligations should be sourcing data at the plant level rather than the country level. Does Indonesian material qualify for incentives in our target market? Check ownership and control, not just origin. This is the single most common and most expensive sourcing error in this market right now. Three ways this resolves The next five years are genuinely uncertain, and the honest framing is scenario rather than forecast. Consolidation. Prices stay low enough that marginal capacity stops running, and the industry concentrates into the lowest-cost operators. Indonesia’s position strengthens on volume while its fiscal take per tonne weakens. This is, on present evidence, the path of least resistance. Escalation. Indonesia moves decisively up the table — cathode, cell, and eventually vehicle assembly, supported by incentives and by requirements attached to domestic market access. This requires technology partnerships and customer qualification that no export rule can mandate. Several projects are aimed at it. Whether they reach commercial scale on schedule is open. Fragmentation. Buyers in the United States and Europe build qualified supply outside Chinese-controlled chains, while Chinese-linked capacity serves the rest of the market. Indonesian volume keeps growing but splits into two pools with different prices and different customers. Elements of this are already visible. These are not mutually exclusive. Consolidation and fragmentation can happen at the same time, to different parts of the industry. Glossary Laterite — The ore type Indonesia mines, requiring different processing from the sulphide ores common elsewhere. RKEF — Rotary kiln–electric furnace, the route to ferronickel and nickel pig iron for stainless steel. HPAL — High-pressure acid leaching, the route to mixed hydroxide precipitate for battery-grade material. MHP — Mixed hydroxide precipitate, the intermediate product that HPAL produces and refiners convert to sulfate. FEOC — Foreign entity of concern, the ownership-and-control test that determines clean-vehicle credit eligibility in the United States. The risks worth pricing Technology lock-in. Much of the built capacity was designed around the ore, the capital, and the customer relationships available at the time it was built. If battery chemistry shifts away from high-nickel formulations, that capacity is less valuable than its construction cost. Environmental liability. Land disturbance and tailings practice are live issues now and will remain so. Expect tightening, and expect the compliance cost to land on producers and then on buyers’ documentation requirements. Concentration. Two industrial parks carry most of the capacity. A single-site disruption — power, labour, logistics, regulation — is a supply-chain event, not a local one. Policy drift. Royalty rates, export levies, domestic processing obligations, and local content requirements have all moved. A model built today should be stress-tested against at least two alternative policy settings. The commodity cycle itself. Nickel is a cyclical commodity with a volatile price history, including a short squeeze in 2022 that remains the clearest illustration of how unstable this market can be. Any project economics built on a spot price rather than a through-cycle assumption should be redone. A checklist before committing capital Establish which row of the chain you are actually investing in, and whether the margin at that row justifies the capital Trace ownership and control of your counterparty to the level required by your target market’s rules — not to the level that is convenient Source emissions data at plant level, and confirm it survives your reporting obligations Model at least two nickel prices and two policy settings; if the project only works at one, it is not a project Confirm power supply, logistics, and port access are contracted rather than assumed Check whether your offtake customer has already qualified the specific plant, or whether that process has not started Ask what happens to your asset if battery chemistry moves away from high-nickel The honest bottom line Indonesia’s nickel policy is neither the triumph its supporters describe nor the cautionary tale its critics prefer. It did the thing it set out to do: it moved processing into the country, created a large industrial base where there was essentially none, and changed the structure of a global commodity market within five years. Very few industrial policies in the developing world have done that much, that fast. What it did not do is the harder part. Banning an export moves the location of an activity. It does not move the location of the margin, the technology, or the customer relationship — those are earned at the top of the chain, and they are earned by qualification, process capability, and access, none of which respond to a prohibition. For a company deciding what to do about it, the practical conclusion is unromantic. Indonesian nickel is abundant, cheap, and increasingly central to the world’s battery supply. It is also concentrated, carbon-intensive, policy-exposed, and tangled in other countries’ ownership rules. Those facts are not contradictory. They are the deal, and the companies that do well here are the ones that priced all of them rather than the ones that found the cheapest tonne.

Christian Brooks – SeaPRwire – Johor Bahru spent most of its modern history as Singapore’s industrial periphery — a place where palm oil estates and light manufacturing absorbed overflow labour from across the strait without ever becoming anything like a destination. Commuters knew it as a traffic jam and a weekend shopping trip. Nobody called it a digital infrastructure market. Then a regulatory decision made eighty kilometres away, in a jurisdiction with no say over Malaysian planning policy, turned it into one of the fastest-growing data centre markets on earth. This is the story of how that happened, what the numbers actually look like in 2026, where the real constraints now sit, and why the answer to “should we build there” is much less obvious than the growth charts suggest. One moratorium redirected an entire industry Between 2019 and 2022, Singapore stopped approving new data centres. The city-state cited the pressure energy-intensive facilities placed on already constrained land, water, and electricity systems. It was a reasonable decision for Singapore, and it did nothing whatsoever to reduce regional demand for compute. Demand went looking. Johor was the only place close enough, cheap enough, and politically willing enough to absorb it. What that meant concretely: Singapore did not just pause approvals, it created a backlog of capital already committed to the region with nowhere to land. Hyperscale cloud providers, colocation operators, and Chinese technology firms seeking a Southeast Asian foothold all hit the same constraint at the same time. Johor offered land at a fraction of Singapore’s price, a government actively courting investment rather than restricting it, and physical proximity close enough to preserve the low-latency connectivity that made Singapore valuable in the first place. None of this happened because Malaysia out-competed Singapore in a contested bid. It happened because Singapore closed a door and Johor was standing on the other side of it with the lights on. The timing mattered enormously. Johor’s expansion began as an overflow market for conventional cloud and enterprise workloads. Then ChatGPT launched in November 2022, and AI-specific demand layered directly on top of spillover demand rather than arriving as a separate wave. An open door, a pre-existing connectivity advantage, and an AI demand curve arriving at exactly the right moment is what produced growth rates no other Asia-Pacific market has matched. The numbers, stated carefully Growth figures in this market get quoted loosely, so here they are with their sources attached. Metric Value Source / date Johor completed capacity ~850 MW JLL, Q2 2026 Under construction ~1.8 GW JLL, Q2 2026 Broader Johor pipeline ~5.8 GW The Diplomat, mid-2025 Five-year capacity growth Over 5,000% JLL regional analysis Implied CAGR since 2020 ~70% JLL regional analysis Malaysia approved DC investment ~USD 26.6bn 2021–2023 cumulative Malaysian DCs operational / building / planned 500+ / ~300 / ~1,140 Ember, 2026 Singapore operational capacity ~1 GW, vacancy 1.4% ARC Group, 2026 Both figures can be true at once: multiple consultancies track this market with different definitions of what counts as “live.” Note also that capacity figures measure electrical supply commissioned for IT load, not necessarily load that is currently drawing power — a distinction that matters enormously for anyone modelling revenue rather than headlines. For context on regional scale, Southeast Asia now accounts for roughly half of all data centre capacity under construction across Asia-Pacific, up from 31 percent at the end of 2025. Malaysia leads with 1,039 MW under construction. Johor alone lifted its under-construction capacity to 602 MW and pushed its total pipeline to 3,088 MW in the first half of 2026 — and Johor retained top spot in Cushman & Wakefield’s regional maturity index across thirty markets. Who is actually building it This is not an organic market assembled by dozens of entrants. It was built fast by a small number of very large operators. As of December 2024, Bridge Data Centers — a subsidiary of Bain Capital-backed ChinData — operated about 32 percent of Malaysia’s built IT capacity, with DayOne at 29 percent and AirTrunk at 13 percent. Three operators, roughly three-quarters of the market. That concentration is typical of hyperscale-led buildouts, and it has consequences: bargaining power, land assembly dynamics, and a market where a handful of decisions explain most observed outcomes. The tenant list reads like a roll call of the AI economy. YTL Power’s Kulai campus, built in partnership with Nvidia, was among the first to break ground at scale in Johor rather than across the causeway. Microsoft has acquired land and planned a Johor cloud region. Oracle committed roughly USD 6.5 billion to a cloud region. Google put USD 2 billion into its first Malaysian data centre in 2024, followed by a sovereign cloud partnership with Dagang NeXchange. Alibaba Cloud opened a Johor cloud region in mid-2026 — its largest digital infrastructure footprint in Southeast Asia to date. Capital markets noticed. Blackstone completed its A$24 billion acquisition of AirTrunk in December 2024. A KKR-led consortium with Singtel agreed in February 2026 to acquire the remaining 82 percent of ST Telemedia Global Data Centres at an enterprise value of S$13.8 billion — the largest transaction in Singapore digital-infrastructure history — taking KKR to 75 percent of a platform with 2.3 GW of design capacity across twelve markets. The strategic sorting: Singapore keeps what it is good at It is tempting to read this as Singapore losing its industry. That reading is wrong, and operators who believe it make bad siting decisions. Singapore is deliberately not trying to be the region’s scale-out location. Since lifting the 2022 pause it has released capacity selectively and conditionally — the IMDA and EDB capacity call launched in December 2025 offered at least 200 MW conditioned on green energy commitments, with applications closing in March 2026. Rationale: latency-sensitive financial and enterprise workloads, hyperscaler regional headquarters, and the most mature regulatory frameworks stay on the island. Raw training capacity goes where power and land are cheap. Microsoft’s USD 5.5 billion Singapore commitment, announced in early 2026 as part of a broader USD 6.5 billion Southeast Asian package, is the largest technology infrastructure investment the country has ever received. Amazon Web Services had already committed USD 9 billion in 2024. These are not exit signals. The clearest illustration is Singtel Nxera’s DC Tuas: opened in early 2026 at 58 MW with a design PUE of 1.23–1.25, large-scale direct-to-chip liquid cooling, an integrated cable landing station, and more than 90 percent of capacity pre-committed before opening. That is a facility designed for governance and interconnect value, not for cheap megawatts. The result is one functional market spanning three jurisdictions — Singapore supplying capital, connectivity, governance and demand; Johor supplying land and power; Batam emerging as the third leg with a 463 MW pipeline and the Firmus/Nvidia/DayOne AI factory campus targeting 360 MW and up to 170,000 GPUs through 2027–28. The Batam GPU farms get hard-wired into Singapore’s ecosystem by the Singtel–Telin INSICA cable with dedicated fibre pairs ready for service in late 2026. Where it is actually breaking Three constraints now bind this market, and they are not the ones a land-cost analysis would surface. Power is first. Grid capacity has not kept pace with commitments. Tenaga Nasional Berhad has told several operators that new substation capacity in parts of the Kulai corridor will not be ready before 2028 — a timeline mismatch now shaping where inside Johor new campuses get sited, pushing some proposals toward Iskandar Puteri where transmission headroom is less committed, and further out to Pontian and Batu Pahat. Malaysia plans up to eight gigawatts of new gas-fired power by 2030 specifically to meet data centre demand, and has revived a nuclear programme with a 2031 target largely because of AI electricity consumption. Water is second. Regulators estimate approved facilities could demand 675 million litres of water a day against roughly 142 million sustainably suppliable. That gap produced an approximately eighteen-month pause on new water-cooled approvals from late 2025. Any operator whose design assumes evaporative cooling needs to have a different answer ready. Cost is third and least discussed. Malaysia’s July 2025 electricity tariff rebasing lifted data centre power costs by an estimated 10 to 14 percent. The cheap power that originally justified the cross-border move is not as cheap as it was, and nobody should model 2023 economics into a 2028 business case. Underneath all three sits a political constraint that has no spreadsheet equivalent: social licence. Malaysia saw its first-ever data centre protest in February 2026, when residents in Gelang Patah rallied against a ZDATA site over dust, pollution, and water concerns. Calls for a broader moratorium grew through mid-2026. Meanwhile the federal government has effectively stopped approving new projects not demonstrably linked to AI — a selective freeze disclosed by the Prime Minister in February 2026 after roughly two years of quiet operation. Community consent is now part of the cost base. The sustainability collision nobody has resolved Here is the tension that will define the next five years. Fossil fuels generate 81 percent of Malaysia’s electricity, with solar and wind providing roughly 2 percent. A standard AI data centre consumes as much electricity as 100,000 households, according to the International Energy Agency. Meanwhile Malaysia has decarbonisation commitments and a MYR 2.1 billion sovereign AI cloud allocation designed to keep data and model training within national borders. These ambitions are on a collision course that has not been resolved. It is not hypothetical — it is visible in the nuclear timeline, in the gas buildout, and in the fact that every new hyperscale approval now carries explicit green energy conditions. For operators, this means the compliance question has shifted. Three years ago the question was whether you could get power. Now it is whether you can credibly document where your power comes from, because hyperscaler tenants increasingly cannot sign contracts that put their own Scope 2 targets at risk. Budget for that documentation as seriously as you budget for substations. A practical siting framework If you are evaluating this corridor, the following questions separate real feasibility studies from optimistic ones. Power before price. Ask Tenaga — or read aloud the answer — what date your specific substation is energised, not when the park promises energisation. Site selection has become an energy question first and a property question second. One operator was told there was no power available before 2028. Cheap land with no energised substation is not cheap. Cooling before capacity. With water-cooled approvals paused, know whether your design works in this climate within available water allocations before you commit to a site. Compliance before construction. Malaysia’s new approval regime rewards demonstrable AI linkage and punishes generic capacity. Applications framed as general colocation are not being approved. Community before concrete. The Gelang Patah protest was not an anomaly. Engage the adjacent community before commencement, not after complaints. Latency honesty. Workloads that genuinely need to be within milliseconds of Singapore’s financial district belong in Singapore. Workloads that need cheap megawatts belong in Johor or Batam. Pretending one site does both is how operators end up paying Singapore rents for a training cluster. Questions people actually ask Is the boom sustainable? The demand side is real and AI-driven. The supply side is now constrained by grid, water, and political consent — the same three factors that constrained Singapore in 2019, which is a useful warning. Will Singapore reclaim the capacity? Partly, selectively, and only for high-value low-latency workloads. It has explicitly stopped competing on raw scale. Can we rely on the developer’s power timeline? Treat any energisation date that does not come from the utility as a marketing estimate. The single most common reason projects slip in this corridor is a gap between when a developer promised power and when the substation was genuinely live. Get the utility’s answer in writing, dated, before you sign anything that carries a delivery obligation. Is it too late to enter? For generic speculative capacity, yes — that window closed with the AI-only approval policy. For differentiated capacity with secured power and a credible sustainability story, there is still room. What about Batam? Worth watching closely. The GPU campus scale and the dedicated cable are serious commitments, and its proximity to Singapore mirrors Johor’s original logic. It is roughly eighteen months behind Johor on the same curve. What is the realistic downside? A grid delay slipping your energisation by two years while your anchor tenant’s contract has a delivery date is the scenario that actually kills projects here, not land cost or permitting fees. Cooling has become a siting decision Cooling used to be an engineering footnote decided after the site was bought. In Johor in 2026 it is a gating constraint, because the eighteen-month pause on new water-cooled approvals removed the default option most designs assumed. The arithmetic is stark: approved facilities could require 675 million litres of water per day against roughly 142 million sustainably suppliable. Even allowing for conservative estimates on the demand side, the gap is roughly fourfold. No amount of engineering tweaks closes a gap of that size; only changing the cooling paradigm does. Practically, operators now choose among three paths, each with a distinct trade-off. Air-cooled designs at lower rack density reduce water to near zero but sacrifice the power density that AI training clusters require. Closed-loop liquid cooling with dry coolers preserves density at higher capital cost and some efficiency penalty in a tropical climate. The third path — the one everyone asks about first — is to wait out the pause, which is a bet that regulation loosens rather than a plan. What makes this a genuine siting decision rather than a design detail is that the answer varies within Johor. Sedenak Tech Park alone hosts over half the state’s live capacity, and its water allocation situation differs from newer proposals directed at Pontian and Batu Phat further from Singapore’s border. Sites further from the causeway trade latency for utility headroom, and for training workloads that trade is usually worth making. What locating here actually costs Published land prices are the least informative number in any Johor feasibility study. The costs that decide projects sit elsewhere. Cost driver What changed recently Effect on feasibility Electricity tariff Rebased July 2025 Data centre power costs up an estimated 10–14% Substation connection Turnk lead times pushed past 2028 in Kulai Can delay revenue by 1–2 years Water allocation New water-cooled approvals paused ~18 months from late 2025 Forces redesigned cooling, higher capex Land in Sedenak / Kulai Applications exceed zoned industrial land Pushes projects to Pontian, Batu Phat Compliance and ESG documentation Now a condition of hyperscaler contracts Adds professional services cost but unlocks tenants The tariff change deserves particular attention because it quietly eroded the original premise. The reason to cross the causeway was cheaper power. Power is still cheaper in Johor than in Singapore, but the margin narrowed by double digits in a single year, and there is no institutional commitment that it will narrow no further. Model both spreads, not just today’s. How Malaysia sits in the regional picture Johor’s position looks very different depending on which comparison you make, and picking the wrong comparator produces bad strategy. Against Singapore, Malaysia wins on land, power availability, and cost, and loses on latency, governance maturity, and connectivity density. That complementarity is exactly why the cross-border model works — it is not a competition, it is one supply chain. Against Thailand, the picture is less comfortable. Bangkok posted Asia-Pacific’s fastest data centre expansion with a 148 percent jump in under-construction capacity, representing the whole of Thailand’s 859 MW of activity. Thailand offers large domestic demand, a stable grid, and no comparable water controversy yet. Operators who assume Johor is the only Southeast Asian option are not looking at the same maps their competitors use. Against Indonesia and Batam, proximity to Singapore matters more than national policy differences. Batam’s dedicated cable and GPU campus commitment mirror Johor’s original logic almost exactly, eighteen months later. If you are making a decision today, Batam deserves evaluation on the same shortlist — its disadvantage is timing, its advantage may be a cleaner slate on grid and water planning. Glossary Energisation date — The date a substation is actually live and able to deliver contracted power, as opposed to the date a park or developer promises availability. IT load — The electrical capacity available to computing equipment, excluding cooling and other facility overhead. PUE (Power Usage Effectiveness) — Total facility energy divided by IT equipment energy. Lower is better; 1.23 is strong, 1.5 is mediocre. Hyperscaler — A cloud provider operating at very large scale, typically the anchor tenant in a wholesale facility. Colocation — A facility leasing space, power, and cooling to multiple tenants rather than to a single anchor. Forward calls, and how they could be wrong Directionally, the corridor consolidates. Malaysia is heading toward 2 GW of operational capacity by 2028, and the strategic geometry — Singapore for inference and governance, Johor and Batam for raw training — looks stable for at least five years. Three things could break it. First, if power timelines slip further while AI capital demands faster deployment, capacity chases reliability into other markets such as Thailand, where Bangkok posted the region’s fastest expansion with a 148 percent jump. Second, if water politics escalate, evaporative cooling becomes untenable and retrofitting costs land on operators who assumed otherwise. Third, if data localisation requirements harden in ways that conflict with cross-border compute patterns, the single-market logic of the corridor fractures. The single most useful thing to track is not headline capacity but substation energisation dates. That is the variable every operator complains about privately and almost nobody publishes. The honest bottom line Johor’s rise was not a development programme. It was the unplanned consequence of one jurisdiction’s constraint meeting another’s willingness, amplified by the largest compute demand shock in a generation. For operators, the practical takeaway is unglamorous: this market rewards those who secure power, water, permits, and community consent in that order, and punishes those who buy land first and discover later that power arrives in 2028. The growth charts are genuinely extraordinary. They are also a lagging indicator — the decision that shaped this market was made in Singapore in 2019, and the decisions shaping the next five years are being made right now in substation planning offices and municipal council meetings that almost no international coverage mentions.

Robert Sterling – SeaPRwire – A hotel in Cebu lost a booking last March because of a QR code. Not a broken one — a missing one. A guest from Singapore stood at the front desk with her banking app open, ready to pay, and there was nothing to scan. She paid in cash from an ATM down the road, complained in a review the same evening, and the property’s operations manager spent the next two weeks explaining to ownership why a “cashless” property needed a cash float again. That is the shape of the problem most Southeast Asian businesses actually have with payments in 2026. It is not fraud, and it is not interchange. It is that the person standing in front of you has already decided how they want to pay, and your checkout was built for a card. This is a practical guide to the rails that matter, what they cost, how long money takes to arrive, where it gets stuck, and which countries quietly make it hard to take your own money out. Cards are not the default here A merchant entering Southeast Asia with Visa and Mastercard acceptance and nothing else has already lost most of the market. Cards account for under 15 percent of potential customers across most ASEAN markets. In Thailand, account-to-account payments make up 44 percent of e-commerce value. In the Philippines, GCash alone has more users than most countries have adults, and GrabPay, GoPay, OVO, DANA, TrueMoney, and Touch ‘n Go each hold meaningful share at home. The region runs instead on government-built real-time rails and QR standards, and they are not “alternative payment methods” in the way buy-now-pay-later is in Europe. They are the default. Research points to local payment methods representing 69 percent of Asia Pacific e-commerce transaction value by 2028, up from 64 percent in 2023, with non-card wallets reaching 57 percent. The trajectory is away from cards, not toward them. The reason is partly cultural, but mostly structural. These rails were built by central banks to bring unbanked populations into the financial system, and they priced merchant acceptance at near zero to make that work. A QR payment costs a merchant far less than a card transaction — estimates run 50 to 80 percent lower in acceptance cost — and that pricing floor is very hard for card networks to match. The five rails that matter Every market built its own. That is the single most expensive fact in this article. Market Real-time rail Operator QR standard Settlement Singapore PayNow (over FAST) ABS / BCS SGQR Near-instant Malaysia DuitNow (RPP) PayNet DuitNow QR Near-instant Thailand PromptPay ITMX Thai QR Near-instant Indonesia BI-FAST Bank Indonesia QRIS Near-instant Vietnam NAPAS 247 NAPAS VietQR Near-instant Philippines InstaPay / PESONet BSP QR Ph Instant / same-day Each has its own integration standard, settlement timing, and local partner ecosystem. An institution offering collections across five markets is not building “an ASEAN offering.” It is building five distinct market capabilities that happen to sit next to each other on the map. The scale behind these rails is larger than most outsiders assume. Indonesia’s QRIS has passed 56 million users and roughly 38 million merchants. Malaysia’s DuitNow QR doubled to 3 billion transactions. PromptPay carries 74 million registered users and processes over 74 million transactions daily. Cross-border QR actually works now The interesting development of the last eighteen months is that these national systems started talking to each other. This is genuinely new, and it changes the calculus for merchants whose customers cross borders. Cross-border QR payments work through interoperable regional networks. A visitor from Singapore opens a PayNow-enabled banking app, scans a Philippine merchant’s QR code, and approves the payment in SGD from her Singapore account. The merchant receives the equivalent in PHP. The merchant does not need a foreign currency account. There is no currency exchange at the point of sale. Settlement timing differs from domestic transactions, and it is the detail most merchants get wrong when modelling cash flow: domestic transactions settle next business day in local currency; cross-border QR transactions take longer — typically T+1 to T+2 calendar days depending on the corridor and provider. Plan for that gap during peak tourist periods. By 2026, cross-border QR was live across Indonesia, Thailand, Singapore, Malaysia, and the Philippines, with more than 30 digital payment schemes supported through shared infrastructure. Singapore’s PayNow also links to India’s UPI. ASEAN had 29 cross-border payment linkages processing 36.2 million transactions worth about USD 716 million in one reported period, and the Regional Payment Connectivity initiative kept adding to that. Project Nexus — the BIS Innovation Hub scheme operated through Nexus Global Payments in Singapore — was on track for rollout in 2026, connecting the instant payment systems of India, Malaysia, the Philippines, Singapore, and Thailand through a central gateway. If that lands as expected, the plumbing gets materially simpler for anyone serving several markets at once. The licensing wall nobody budgets for Here is where optimistic expansion plans usually die. Each market regulates money movement separately, and no regional passport exists. Market Regulator Representative licence Indicative timeline Singapore MAS Major Payment Institution ~2 years Malaysia Bank Negara Malaysia Money Services Business 2+ years Philippines BSP Remittance & Transfer Company / EMI ~1 year Indonesia Bank Indonesia Payment Service Provider (PJP) ~2 years Thailand Bank of Thailand / MoF Designated Payment Service ~3 years Vietnam State Bank of Vietnam Intermediary Payment Service 3+ years A licence held in Singapore does not extend to IDR flows in Indonesia or VND collections in Vietnam. For most businesses that means either pursuing local licensing market by market — a multi-year undertaking — or working through local partners, each adding counterparty risk, commercial negotiation, and operational oversight to the stack. The practical consequence: unless you are a payments company, do not try to hold licences. Use a licensed local provider in each market and accept that you are assembling a patchwork rather than buying a platform. The trap nobody warns you about: getting your own money out This is the section most guides skip, and it is where operators lose real money. Some ASEAN currencies carry capital controls, restricted convertibility, or repatriation rules that do not apply to SGD or USD. The Indonesian rupiah, the Vietnamese dong, and to a lesser extent the Philippine peso and Thai baht all sit somewhere on that spectrum. Funds collected locally cannot always be freely converted and moved out. Market Convertibility What it means operationally Singapore Fully convertible Standard FX hedging available, freely deliverable Malaysia Not internationalised Offshore MYR trading not recognised; FX must transact onshore Philippines Largely convertible (current account) Deliverable onshore, often hedged via NDF Indonesia Heavily controlled Onshore rails plus documentation trail for every conversion Vietnam Non-convertible offshore VND cannot be traded outside Vietnam; FX needs documented purpose For a business running treasury across several markets, this creates genuine balance-sheet exposure: cash trapped onshore, unpredictable conversion timing, and pressure to accept unfavourable rates to unwind positions quickly. Model this before you open the second market, not after. Choosing a provider, honestly The honest answer is that no single platform covers all of ASEAN well, and any vendor that claims otherwise is describing aspirational coverage rather than live capability. What you are really choosing is how much fragmentation you will manage yourself. The trade-offs worth weighing explicitly: local payment method depth versus breadth of markets; whether the provider holds its own licence or rides on a partner’s; settlement speed and whether it varies by rail; what happens when a transaction fails; and whether reconciliation actually reconciles, which is where operators lose the most staff hours. A mid-sized regional business typically ends up with two or three providers rather than one. That is not a failure of planning. It is the current shape of the market. The mistake is assuming you can consolidate later without redoing integrations. Provider type Strength Watch out for Regional PSP with broad local methods Fastest route to real coverage Settlement varies by corridor; check per-rail terms Card-first global gateway Familiar integration, strong fraud tooling Thin local rail coverage; poor non-card conversion Orchestration layer One API across many providers Adds cost; still needs underlying local licences Local bank acquiring Best pricing on that rail One market only; slow to extend How this differs country by country Country-level differences are large enough that a single regional playbook will produce bad decisions. Indonesia — QRIS is the gravity well, with 56 million-plus users. But IDR is heavily controlled, so plan repatriation before you plan acquisition. Documentation discipline matters more here than anywhere else in the region. Philippines — Strong English capability makes integration easier, but the payment mix is fragmented across GCash, Maya, QR Ph, InstaPay, and PESONet, plus over-the-counter networks that matter enormously outside Metro Manila. Tourism-facing businesses should treat cross-border QR as baseline, not premium. Thailand — PromptPay is deeply embedded, and tourists from neighbouring markets increasingly expect to pay through linked rails. Watch household-debt-driven softness in domestic consumption when forecasting. Vietnam — VietQR is growing quickly, but VND offshore restrictions are the binding constraint. Build the compliance path first. Malaysia — DuitNow QR doubled to 3 billion transactions. MYR must settle onshore, which shapes treasury design more than most operators expect. Singapore — The most mature and most expensive. PayNow is table stakes, but Singapore is usually where you incorporate and bank, not where your volume sits. Mistakes that cost money Assuming card-first economics still apply. They do not. Budget as though cards are a fallback, because that is what they are for most of your customers. Modelling cross-border settlement as same-day. It rarely is. That gap shows up as working capital. Treating one licence as regional. It is not. Every additional market resets the clock. Ignoring reconciliation until month three. Manual reconciliation across five rails with different reference formats consumes finance headcount faster than any other operational issue here. Discovering repatriation rules after collecting the money. This is the expensive one. Questions businesses actually ask Do I need a local entity to accept payments? Usually not to accept, but often yes to settle directly or to hold a licence. Working through a licensed local provider sidesteps this for most merchants. Is cross-border QR worth enabling for a small business? If your customers include travellers or overseas buyers, yes — the incremental cost is low and the alternative is losing the sale entirely. The service businesses, tour operators, and boutiques that benefit most are the ones with regular foreign footfall. Will Project Nexus remove the fragmentation? Not entirely. It meaningfully reduces integration complexity across connected markets, but licences, currency rules, and reconciliation remain local problems. How long before this is genuinely simple? Realistically, several years at best, and probably not completely. What the money actually costs, line by line Most operators budget for a headline transaction fee and get surprised by everything around it. The real cost of accepting a payment in this region has four layers, and only the first appears on the pricing page. Acceptance cost. This is the merchant discount rate, and it is where local rails win outright. QR acceptance typically runs far below card interchange — the central-bank mandate to serve small merchants with cheap acceptance is why adoption happened at all. Expect QR and real-time transfer costs to sit well under 1 percent at most volumes, against card rates that can run several times that once cross-border and rewards cards are mixed in. FX spread. If a customer pays in SGD and you settle in PHP, someone converts that money, and the spread is rarely the interbank rate. This is usually the largest invisible cost in a cross-border setup, and it is the one most often quoted as “free” in marketing. Always ask for the spread expressed as a percentage above mid-market, not just whether conversion is available. Settlement float. Money in transit is money unavailable. If cross-border QR settles T+2 rather than next business day, a business turning over PHP 3 million a month is carrying roughly two days of working capital at all times. That has a financing cost, even if nobody books it as one. Reconciliation labour. This is the one that quietly dominates for smaller operators. Each rail generates references in a different format, settles on a different schedule, and reports failures differently. A finance officer spending six hours a month per market matching payments is a real cost line, and it scales linearly with the number of rails you accept. The arithmetic that matters: a merchant paying 0.8 percent on a local rail plus a 1.5 percent FX spread plus two days of float is often paying less than a merchant paying 3 percent on a card with instant settlement — but only if volume is high enough for the float to be forecastable. Below a certain monthly turnover, the slower rail is genuinely worse for cash flow regardless of headline pricing. A worked example: the tour operator’s cash cycle Consider a dive operator in Davao taking bookings from Singaporean, Malaysian, and domestic visitors. Before enabling cross-border QR, roughly 40 percent of foreign enquiries that reached the payment stage dropped out — the booking form offered card or bank transfer, and neither matched how the customer habitually paid. Of those lost bookings, the operator estimated that about half were recoverable if the visitor could pay from their home app. After enabling PayNow, DuitNow, and QRIS acceptance alongside QR Ph for domestic guests, recovery improved, but the cash cycle changed shape. Domestic bookings settled next business day. Cross-border bookings settled on a T+2 calendar-day cycle. For a business whose deposits fund equipment rental and boat fuel before the guest arrives, that two-day gap mattered more than the fee difference — particularly during Chinese New Year and summer peaks, when a week of concentrated bookings could leave several hundred thousand pesos in transit simultaneously. The lesson generalises: enabling more rails raises conversion and lengthens your cash conversion cycle at the same time. Both effects are real. The mistake is counting the first and discovering the second during your busiest month. When cross-border QR is not worth enabling It is not universally the right move, and saying otherwise does merchants a disservice. Skip or defer it when your customer base is overwhelmingly domestic, when your average ticket is large enough that customers will accept a card or transfer anyway, or when you lack the staff to handle reconciliation across additional rails. A B2B manufacturer invoicing three buyers twice a month does not need five payment methods; it needs reliable settlement on one. The businesses that clearly benefit are those with regular foreign footfall — tourism-facing retail and F&B in places like BGC, Intramuros, Boracay, and Cebu; export-linked e-commerce selling to Southeast Asian buyers; event and trade-show operators; and service businesses invoicing overseas clients who hold no local bank account. For those, acceptance is increasingly a baseline requirement rather than a premium feature. Which rail to launch first Teams routinely ask which single market to integrate first, and the answer is less about market size than about what the first integration teaches you. Start in the market where your own entity already sits, because licence questions, bank onboarding, and FX behaviour are all easier where you already have local standing. Your first integration will surface every gap in your reconciliation tooling, and it is far better to find those gaps somewhere you already understand the language, the banking hours, and the regulator. Once that first rail is stable, the sequencing that usually works is: add the market sharing the most customers with the first, then the market whose foreign tourists most often appear in your funnel, then everything else. What rarely works is trying to launch three markets simultaneously because regional coverage looks achievable on a provider’s coverage map. Every integration you run in parallel multiplies the number of failure modes you have not yet diagnosed, and it is much harder to tell whether a problem lives in your integration, the provider, or the local rail. One practical rule worth adopting: do not add a second rail until the first one has run for a full month with clean reconciliation. A single clean month tells you whether your reference formats, failure handling, and payout timing are genuinely working, and that knowledge transfers to the next market with very little adjustment. Glossary Real-time rail — A national instant payment network operated or overseen by the central bank, moving funds between accounts in seconds. QR interoperability — The arrangement that lets a customer scan a merchant QR using a wallet or banking app from a different country. NDF — Non-deliverable forward, a cash-settled FX derivative used to hedge currencies that cannot be freely delivered offshore. T+2 — Settlement two calendar days after the transaction, as opposed to next business day. PJP / MPI / MSB — Local payment licence categories in Indonesia, Singapore, and Malaysia respectively. What changes next, and what could go wrong The direction of travel is toward fewer, better-connected rails. Southeast Asia’s digital payments already crossed USD 1 trillion in transaction value. National QR systems keep growing, and cross-border linkages keep multiplying. The risks run the other way too. Data localisation requirements are tightening in several markets, which could push against harmonisation. Geopolitical friction affects corridors that currently work smoothly. And the same financial-inclusion mandate that made these rails cheap to begin with could make regulators reluctant to let them become the plumbing for cross-border capital flows. The practical stance for 2026: build for fragmentation, buy local depth, and keep any single-market integration small enough that unwinding it is survivable. A checklist you can run this week List the five payment methods your last 100 customers actually tried to use Check whether your current provider settles each rail at the same speed — read the per-rail terms, not the headline Ask your provider directly: which licence do you hold, and whose do you ride on? Model repatriation for your largest non-SGD, non-USD exposure before adding another market Test one cross-border QR transaction end to end with a colleague in another country Count the staff hours spent on manual reconciliation last month The honest bottom line The businesses that get this right are not the ones that found a magical all-in-one provider. They are the ones that accepted fragmentation as permanent, built a small set of well-understood integrations instead of one fragile universal one, and got religion about reconciliation and repatriation early. Getting paid in Southeast Asia in 2026 is genuinely easier than it was three years ago. It is still five problems wearing one regional label, and treating it that way is the difference between money arriving and money sitting somewhere you cannot reach.