Generalist Hits $3B Valuation With $200M Round Extension

Fundraise Insider tracks newly funded startups each week and delivers verified sales leads of C-suite contacts straight to your inbox, so you reach the right people while the funding is still fresh. See pricing.

Generalist AI has raised roughly $200 million in an extension led by 8VC, joined by existing investors who were not named. The money is an addition to the $400 million Series B that Radical Ventures led in June at a $2 billion valuation, which takes the round to $600 million in total and the company to a $3 billion valuation.

The company works from San Mateo, California, and was founded in 2024 by former Google DeepMind researchers Pete Florence and Andy Zeng together with Andrew Barry, previously an engineer at Boston Dynamics. Nvidia and Bezos Expeditions were among more than half a dozen participants in the June round.

Generalist does not build robots. It builds the foundation models that let existing machines pick up new tasks in unfamiliar settings, which is a different business from selling arms and torsos.

The raise landed about a week after the company introduced Gen-1.5, a model for robotic arms that it says cuts the time needed to build factory automation workflows. The status quo it is aiming at is tedious: developers historically programmed a robotic arm by hand for each task, then updated that code whenever anything changed. Switch a packing line to larger boxes and the machine placing merchandise into them needs revising. Some robots ship with models that reduce custom code, but teaching those systems a new job usually means fine tuning a neural network, which is its own slow project.

Gen-1.5 takes a different route. A person demonstrates the task with their hands while the robot’s built in cameras, or sensors worn on the hands, record what happens. The model can also learn from clips of simulated robots, and according to TechCrunch reporting it can pick up new jobs from demonstrations lasting as little as three to 12 seconds.

Across 10 sample tasks, the company measured an average completion rate of 59% after a single example, rising to 83% once users supplied a few more. Generalist describes it as the first AI model able to learn a broad range of robotics tasks from one or a handful of demonstrations, and says the system refines its own workflows without prompting. During testing it opted to finish some tasks with a different tool than the one it had been told to use.

Neither report specified how the capital will be deployed, though training costs are clearly part of the picture. Gen-1.5 took more than eight months to train, and Nvidia’s presence on the cap table points toward its hardware, including the Jetson line built for robots.

The competitive field is expensive. Physical Intelligence has been valued at around $11 billion, SoftBank-backed Skild AI at $14 billion, and Genesis AI was in talks last month to raise at $3 billion.

Firms selling into newly funded AI and robotics companies can reach executives early through B2B software sales leads built from recent rounds.