The Physical AI startup was acknowledged by judges from NVIDIA, OpenAI, and leading venture firms for its robotics foundation model that has already been deployed in early industrial applications.
MOUNTAIN VIEW, Calif., Dec. 17, 2025 — A Physical AI startup that builds robotics foundation models for real – world deployment has won first place in the Foundation Models, Robot Brains and Runtime category at the 2025 Nebius Robotics & Physical AI Awards, a global competition that recognizes early – stage companies advancing next – generation robotics and Physical AI.
Hosted by Nebius (NASDAQ: NBIS) and supported by NVIDIA, the awards were announced at the Computer History Museum in Mountain View as part of the inaugural Nebius Robotics & Physical AI Awards and Summit. The program gave $1.5 million in AI cloud compute and inference credits to startups selected from more than 250 applicants across 60+ countries.
RLWRLD was chosen as the top winner after evaluations by a judging panel that included leaders from NVIDIA, OpenAI, ABB, Physical Intelligence, and venture firms such as Khosla Ventures, Accel, Next47, and Shanda Grab Ventures. Judges evaluated finalists based on technical originality, scalability, and readiness for real – world deployment.
As the first – place winner in the Foundation Models category, RLWRLD received $150,000 in Nebius AI Cloud compute and inference credits, which are accelerated by NVIDIA AI infrastructure.
The Nebius Robotics & Physical AI Awards were established to highlight startups building the models and systems needed to deploy Physical AI in real industrial settings. Sixteen companies were recognized across categories including Foundation Models, Simulation, Robot Runtime, Industrial Deployment, and Vision AI.
RLWRLD’s work centers on building robot foundation models designed for high – difficulty manipulation and reliable industrial deployment, rather than technology demonstrations for general – purpose robotics. At the heart of its platform is RLDX, a proprietary robotics foundation model capable of controlling a 15 – degree – of – freedom robotic hand, allowing for precise manipulation tasks that demand human – level dexterity.
Judges also mentioned RLWRLD’s use of high – fidelity, multimodal industrial data and its active cooperation with industry partners as signs of the company’s commercial readiness, especially for manufacturing and logistics environments where reliability and precision are crucial.
“Winning this category strengthens our belief that Physical AI needs foundation models built for the real world, not just controlled demonstrations,” said Jung – hee Ryu, CEO of RLWRLD. “This recognition shows the progress we’ve made by working directly with industry partners to deploy our technology.”
After receiving the award, RLWRLD plans to expand global partnerships and speed up the commercial deployment of its foundation models across humanoid and industrial robotics, with an emphasis on manufacturing and logistics environments in the U.S. and Asia.
About RLWRLD
Founded in 2024, RLWRLD is a Physical AI startup developing robot foundation models that enable machines to see, sense, and act with human – like dexterity and cognition. The company operates in the United States, South Korea, and Japan, and trains its models using high – precision 4D+ multimodal industrial data collected through a proprietary multi – sensor capture system. RLWRLD is currently collaborating with major manufacturing partners in South Korea and Japan and aims to become a global leader in industrial robotics AI.
Media Contact
Kite Hill
RLWRLD@kitehillpr.com
