Nebius and Nvidia Launch Physical AI Living Lab to Accelerate Europe's Robotics Startups
One of the biggest challenges in robotics today is no longer building a robot. It is building the infrastructure needed to train one.
While advances in AI have dramatically improved robot perception, planning, and autonomy, many robotics startups still face the same bottlenecks: access to simulation environments, synthetic data generation, and the massive computing resources required to train increasingly sophisticated models.
Nebius believes it can help solve that problem.
The cloud infrastructure provider has launched the Physical AI Living Lab, a six-month program designed to give robotics startups across the United Kingdom and Europe access to Nvidia's growing ecosystem of physical AI tools. The initiative combines cloud computing resources, simulation software, synthetic data generation, and technical support into a single package aimed at helping young companies move more quickly from prototype to deployment.
The first cohort is expected to begin in September through the Nvidia Inception startup program.
The Infrastructure Problem Behind Physical AI
The robotics industry often celebrates breakthroughs in hardware and AI models, but the infrastructure behind those advances receives far less attention.
Training a modern robot increasingly requires more than collecting real-world data. Developers need large-scale simulation environments, synthetic training datasets, digital twins, and computing clusters capable of processing enormous amounts of information.
For many startups, assembling that stack can be as challenging as developing the robot itself.
"Most robotics teams can build a strong model—the bottleneck is getting the simulation, synthetic data, and compute in place to take it further," said Evan Helda, Head of Physical AI at Nebius.
The Living Lab is designed to address exactly that issue by providing startups with access to Nvidia's physical AI ecosystem running on Nebius AI Cloud infrastructure.
Participants will gain access to technologies including Nvidia Cosmos world foundation models, Isaac simulation tools, and synthetic data generation capabilities that can help accelerate robot development while reducing dependence on costly real-world testing.
Why Simulation Matters More Than Ever
The announcement reflects a larger shift taking place across robotics.
Historically, robots learned primarily through direct interaction with the physical world. While that approach remains important, it is expensive, slow, and often difficult to scale.
Physical AI developers are increasingly relying on simulation-first approaches that allow robots to practice millions of scenarios before entering real environments. Synthetic data can expose systems to edge cases that might rarely occur during real-world testing, while digital twins allow developers to evaluate performance before deploying hardware.
The result is a development process that increasingly resembles modern software engineering rather than traditional robotics.
As foundation models move into robotics and embodied AI systems become more general-purpose, simulation and synthetic data are becoming critical pieces of the development pipeline.
Strengthening Europe's Physical AI Ecosystem
The launch also highlights growing efforts to strengthen Europe's position in the emerging physical AI economy.
The United Kingdom has become a significant center for robotics research, producing world-class work from institutions including Imperial College London, Oxford, Cambridge, Bristol, and Edinburgh. Yet many observers have noted a persistent gap between academic innovation and commercial deployment.
Anthony Hills, Nvidia's Director for the UK and Ireland, sees infrastructure as one of the missing pieces.
"The UK has world-class robotics and AI research, but there's still a real gap between that innovation and scaled, market-ready solutions in physical AI," Hills said.
By reducing barriers associated with computing resources and development tools, the program aims to help startups bridge the transition from laboratory prototypes to commercial systems.
More Than a Startup Program
For Nebius, the Living Lab is also a strategic opportunity.
The company is positioning itself as a key infrastructure provider for the next generation of physical AI companies. By working directly with robotics startups, Nebius gains insight into the computational and software requirements of emerging applications ranging from warehouse automation and industrial robotics to autonomous systems and humanoids.
Those lessons will help shape future iterations of the program and inform planned expansions into additional regions.
The initiative also deepens Nebius' relationship with Nvidia, whose software ecosystem is rapidly becoming the default development platform for many physical AI companies.
Building the Physical AI Stack
The announcement underscores an increasingly important reality: the future of robotics will depend as much on infrastructure as on hardware.
The winners in physical AI may not simply be the companies building robots. They may also include the organizations providing the simulation environments, cloud computing resources, synthetic data pipelines, and development platforms that make those robots possible.
Programs like the Physical AI Living Lab suggest that the race to build embodied AI is becoming a race to build the ecosystem around it.
And for Europe's robotics startups, access to that ecosystem may prove just as valuable as access to capital.
