General Intuition Eyes $6 Billion Valuation as Humanoid Brain Race Accelerates

The race to build the cognitive brains for the next generation of humanoid robots is reaching eye-watering financial heights. General Intuition, a startup developing foundation models to train generalized AI agents to navigate physical space, is reportedly in talks to secure fresh funding at a staggering $6 billion pre-money valuation. The round is poised to draw backing from prominent venture firms including Valor Ventures, Point72 Ventures, and Seven Seven Six, signaling deep institutional confidence in the software layer of the robotics revolution.

This multi-billion-dollar valuation highlights a critical shift in the humanoid robotics market. While hardware manufacturers focus on refining bipedal frames like Figure 03, Optimus, and Apollo, the ultimate utility of these machines hinges on physical AI. General Intuition’s models are designed to teach robots how to interact with unpredictable environments, translate sensory inputs into physical actions, and master complex manipulation tasks. By decoupling the brain from the physical chassis, software developers aim to create universal operating systems that can be licensed across various humanoid platforms.

From a business desk perspective, a $6 billion valuation for a physical AI startup represents an immense bet on future licensing revenue and ecosystem dominance, especially considering that widespread commercial deployments of humanoids remain in their infancy. This funding momentum coincides with broader consolidation in the physical AI sector, highlighted by enterprise data player EXL recently acquiring physical AI model developer iMerit. Investors are betting that the software layer will yield much higher margins than capital-intensive hardware manufacturing. However, these valuations outrun current commercial realities, where actual unit deployments of catalog humanoids like the Unitree G1 or Digit are still largely confined to pilot programs.

As General Intuition scales its model training, it will face stiff competition from both in-house AI teams at major hardware makers and rival physical AI developers. The success of these foundation models will be judged not by simulation benchmarks, but by how effectively they can be deployed onto physical hardware to perform real-world labor. For humanoid manufacturers, the availability of powerful, off-the-shelf physical AI models could democratize the market, allowing smaller hardware startups to compete with heavily funded giants by focusing solely on mechanical execution rather than proprietary AI research.