Startups & Business News

Genesis AI Raises $105M to Build the Universal Robotics Foundation Model: Why This Matters for the Future of Physical AI

futureTEKnow
July 2, 2025

The robotics world just got a seismic jolt. Genesis AI, a startup founded by Zhou Xian (Carnegie Mellon PhD) and Théophile Gervet (ex-Mistral AI), has emerged from stealth with a massive $105 million seed round co-led by Eclipse Ventures and Khosla Ventures. Their mission? To build a universal robotics foundation model (RFM) and a horizontal platform for general-purpose physical AI—a move that could redefine how machines interact with the real world.

Why Robotics Needs a Universal Foundation Model

Physical labor accounts for a staggering share of global GDP—estimated at $30-40 trillion—but remains 95% unautomated. The bottleneck? Today’s robots are mostly locked into rigid, overfitted software stacks. Industrial arms, for example, are expensive to deploy, hard to scale, and can’t adapt to new tasks without major reprogramming.

Genesis AI wants to break this cycle by developing a foundational model for robots, similar in spirit to what large language models (LLMs) have done for text and language. The goal: create a base model that can power a wide range of robots—across industries and use cases—without the need for task-specific retraining.

The Data Problem: Synthetic Physics to the Rescue

Training AI for robotics isn’t like training an LLM. Robots need to understand and predict the physical world, which means collecting massive amounts of real-world data—a process that’s slow, expensive, and often impractical at scale. Genesis AI’s answer is a proprietary physics simulation engine that generates high-fidelity synthetic data, allowing them to train models faster and more flexibly than competitors relying on off-the-shelf tools like NVIDIA’s software.

This simulation stack, originally developed in collaboration with 18 universities, produces synthetic data at scale and is paired with a real-world data collection system. The result is a closed-loop, data-centric approach that bridges the gap between simulation and reality, giving Genesis AI a unique edge in both speed and data diversity.

What Sets Genesis AI Apart

Full-stack approach: Genesis AI is not just building models; it’s developing the entire data engine, from simulation to real-world collection, to ensure robustness and flexibility.
Open-source ambition: The company plans to open-source parts of its technology, aiming to accelerate progress across the robotics community.
Team pedigree: The founding team brings together talent from Mistral AI, NVIDIA, Google, MIT, and Stanford—an impressive cross-section of AI and robotics expertise.

What’s Next?

Genesis AI’s universal robotics foundation model is slated for release to the broader robotics community by the end of 2025. If successful, this could be the catalyst that finally unlocks scalable, adaptable, and cost-effective automation for physical labor—impacting industries from logistics and manufacturing to healthcare and beyond.

The robotics space is crowded with ambitious startups, but Genesis AI’s data-centric, simulation-first strategy and heavyweight backing make it one to watch as the race for general-purpose physical AI heats up.

futureTEKnow is a leading source for Technology, Startups, and Business News, spotlighting the most innovative companies and breakthrough trends in emerging tech sectors like Artificial Intelligence (AI), Robotics, and the Space Industry.

Become a contributor

Find out how

Discover the companies and startups shaping tomorrow — explore the future of technology today.