Generalist AI, a San Francisco startup founded by researchers from Google DeepMind and Boston Dynamics, just raised $400 million to build the artificial intelligence that could power every robot on Earth, regardless of its physical form. The company's latest model, called GEN-1, demonstrates 99% reliability across diverse dexterous tasks and executes movements up to three times faster than previous state-of-the-art systems.

The funding round, led by Radical Ventures, brings Generalist AI's total raised to more than $500 million. This represents a significant bet on a specific vision of the robotics future: that the real bottleneck in automation is not building better hardware, but creating intelligent software that can understand and act in the physical world regardless of what body it inhabits.

Who's Behind This Robot Brain Company?

Generalist AI was founded in 2024 by three researchers with deep roots in the world's most advanced robotics labs. Pete Florence, the CEO, spent years at Google DeepMind leading development of PaLM-E and RT-2, two of the most influential embodied AI models ever published. His research has been cited more than 19,000 times by other scientists. Andy Zeng, the Chief Scientist, co-authored PaLM-E alongside Florence and previously worked on scaling ChatGPT at OpenAI. Andrew Barry, the CTO, spent five years as a Senior Roboticist at Boston Dynamics, where he helped build Atlas, Spot, and Stretch before transitioning to machine learning research at the Broad Institute of MIT and Harvard.

The investor list reads like a who's who of technology and venture capital. New backers include NVIDIA's NVentures, Bezos Expeditions, and prominent angel investors such as Fei-Fei Li, the founder of World Labs and creator of ImageNet, a foundational dataset for computer vision. Bin Lin, co-founder of Xiaomi, and Naval Ravikant also joined as angels.

What Makes GEN-1 Different From Other Robot AI Models?

Generalist AI's approach differs fundamentally from competitors in the crowded robotics AI space. Rather than optimizing for a single robot or deployment context, the company built GEN-1 to be hardware-agnostic, meaning it works across robotic arms, mobile robots, humanoids, and autonomous systems. This platform approach represents a bet that a single, unified model can be more powerful than specialized systems built for specific machines.

The performance numbers are striking. GEN-1 achieves 99% reliability across diverse dexterous tasks, compared to around 64% reliability for prior state-of-the-art models. The model executes tasks up to three times faster than previous systems and demonstrates what the company calls "emergent improvisational intelligence," the ability to solve physical problems it was never explicitly trained on. It also learns complex new physical skills from limited data, a capability that could dramatically reduce the time and cost needed to deploy robots to new tasks.

The company's competitive advantage rests partly on proprietary data collection methods. According to Boldstart Ventures, Generalist developed a technique using instrumented gloves that capture human manipulation data at scale, creating a training dataset that competitors would find difficult to replicate.

How Does Generalist AI's Strategy Compare to Other Robot Companies?

The robotics AI landscape is splitting into two distinct camps. Some companies, like Figure AI, take a hardware-first approach, vertically integrating their own humanoid robots with proprietary intelligence. Others, like Generalist AI and Physical Intelligence, focus on building foundation models that can power any robot. Physical Intelligence, another San Francisco-based foundation model lab founded by researchers from Google DeepMind, UC Berkeley, and Stanford, is reportedly in talks to raise $1 billion at an $11 billion valuation. Where Physical Intelligence uses a diffusion-based architecture, Generalist emphasizes real-world data at unprecedented scale and a unified model that transfers across hardware form factors.

The unanswered question is which layer captures more value when physical AI scales. Will the companies that own the hardware dominate, or will the companies that own the intelligence layer become the default choice for every robot manufacturer? Figure AI, valued at $39 billion, is betting on hardware integration. Genesis AI raised $105 million for a robotics simulation and foundation model platform, taking yet another approach.

Steps to Understanding the Robot Intelligence Market

• Market Growth Trajectory: The global AI in robotics market is projected to grow at a compound annual growth rate of over 30% through 2033, with service robots growing at over 70% annually in 2025, according to Grand View Research.
• Humanoid Robot Opportunity: Goldman Sachs projects the humanoid robot market alone will reach $38 billion by 2035, with the intelligence layer representing the highest-margin component of that entire stack.
• Scaling Laws in Robotics: In November 2025, Generalist launched GEN-0, the first robotics model to demonstrate scaling laws in the physical world, proving that more physical experience and larger models predictably produce more capable systems.

"Scaling robot learning creates better models, better models can do more useful physical work, and data from real businesses drives the next generation of more capable models. This is how general intelligence will emerge in the physical world: through systems that learn by acting, improve through experience, and become useful by working alongside people," stated Generalist AI in a blog post.

Generalist AI

The defining challenge for Generalist AI is not whether foundation models will power the next generation of robots. GEN-1 makes a compelling case they already do. The real question is whether a company building the brain for every robot can establish itself as the default choice before hardware giants like Figure, Tesla, and Boston Dynamics decide that owning the intelligence layer is too strategically important to outsource.

As the robotics industry matures and moves from research labs into real factories and warehouses, the competition between hardware-first and intelligence-first companies will likely intensify. Generalist AI's $500 million in total funding and the backing of some of technology's most respected researchers suggest the market believes the intelligence layer is where the real value lies.