New York-based General Intuition is in talks to raise fresh funding at a $6 billion pre-money valuation, according to TechCrunch, in a deal that would make the young AI company one of the most closely watched startups in physical AI.

The proposed round would include new investors Valor Equity Partners, Point72 Ventures, and Seven Seven Six, while existing backers Khosla Ventures and General Catalyst are also expected to participate. The financing is not yet final, and TechCrunch reported that the round remains in progress.

The potential valuation would come only weeks after General Intuition raised $320 million at a $2.3 billion valuation. If completed, the sharp increase would signal continued investor demand for companies attempting to connect general-purpose AI models with action in the physical world, particularly robotics.

A foundation model for movement and action

General Intuition is developing what it describes as a foundation model for generalized AI agents that can learn to move through space and time. Its stated direction differs from language-focused systems because the intended output is not only text or images, but behavior: choosing actions, responding to environments, and transferring skills across tasks.

Chief Executive Pim de Witte spun out the company from Medal, his video game clip-sharing platform, in October. According to TechCrunch, General Intuition began with access to hundreds of millions of hours of gameplay and the associated “action labels”—records of which buttons players pressed and when.

That data offers a distinctive starting point for training models around sequences of decisions. Gameplay can provide large volumes of examples involving timing, navigation, objectives, and responses to changing environments. It is not, however, the same as operating a robot in the real world. Translating screen-based actions into reliable physical behavior remains one of the central technical challenges for the company and the broader robotics field.

General Intuition is reportedly using the phrase “large action models” for its approach. The goal is to build a model that can generalize beyond the specific situations represented in its training material rather than simply reproduce known actions.

Investors are pricing in a robotics transition

The reported participation of Valor Equity Partners is notable because the firm is known for backing SpaceX. TechCrunch said a General Intuition investment would be Valor’s first AI lab investment since SpaceX, although the firm had not confirmed the deal when the report was published.

Point72 Ventures and Seven Seven Six would add further new institutional backing, while Khosla Ventures and General Catalyst would provide continuity from the company’s earlier financing. The investor mix reflects a combination of venture interest in foundational AI and a broader appetite for businesses tied to robotics, autonomy, and industrial applications.

TechCrunch reported that the round is oversubscribed, citing a source close to the deal. That is an unverified market signal rather than a completed financing announcement. The final investor list, terms, and amount could still change before closing.

The proposed $6 billion pre-money valuation also needs to be read in context. General Intuition has raised substantial capital in a short period, but the available evidence does not establish commercial revenue, deployed robots, production contracts, or independently measured performance. The valuation is therefore primarily a signal about investor expectations for the company’s technology and market position, not proof that its model has solved physical-world reliability.

What the funding would pay for

According to TechCrunch, General Intuition intends to use the new capital to improve its general model while focusing on robotic embodiments. That plan would require investment in both compute and people.

The company has a partnership with CoreWeave, which TechCrunch identified as part of its compute infrastructure strategy. More funding could allow General Intuition to run larger training experiments, process additional action data, and develop systems capable of adapting across different robotic platforms. It also plans to hire more talent.

For builders, the important distinction is between a model that predicts actions in an abstract environment and one that can control a real machine safely. Robotics systems must contend with sensor noise, latency, mechanical variation, limited battery life, unexpected obstacles, and the cost of failure. A model trained on gameplay may help with temporal reasoning or action selection, but it cannot by itself demonstrate reliable manipulation, navigation, or physical safety.

That creates a deployment question for potential enterprise buyers. A general model could reduce the need to program every robotic task independently, but buyers would still need evaluation procedures, simulation, hardware-specific adaptation, monitoring, and fallback controls. The commercial value will depend on whether General Intuition can make those integration costs lower than the cost of building specialized systems.

Evidence behind the “intuition” thesis

Khosla Ventures investor Vinod Khosla recently told TechCrunch that action labels could contribute to what he called the “emergence of intuition”—the ability of a model to generalize to tasks it was not explicitly trained to perform. That is an investor view, not an independently validated result from General Intuition.

The company’s underlying evidence base, as described in the report, is its gameplay-derived dataset and its ongoing model development. The source does not provide independent benchmark results, a public technical paper, customer deployments, or details on how performance is measured across real-world robotic tasks.

That gap matters because generalization claims are difficult to assess in robotics. A system may perform well in a controlled simulation or on a narrow family of tasks while failing when the environment, object, hardware, or objective changes. Investors may be willing to fund that research risk, but enterprise adoption will require reproducible tests that connect model performance to operational outcomes.

What to watch next

The first signal will be whether General Intuition closes the reported round and confirms its valuation, size, and investor roster. Valor Equity Partners’ involvement is also worth watching because the firm had not publicly confirmed its participation at the time of TechCrunch’s report.

Technical disclosures will be more important than the headline valuation. Useful evidence would include evaluations of the company’s large action models on robotic embodiments, comparisons between gameplay pretraining and robot-specific data, and results showing transfer to tasks or environments that were not part of training.

Product teams and researchers should also watch for details on the CoreWeave partnership, the kinds of robots General Intuition supports, and whether the company offers a platform for developers or remains focused on internal model research. Customer pilots, safety practices, and hardware integration requirements would provide a clearer view of how close the company is to practical deployment.

Creati.ai perspective

General Intuition’s reported financing highlights a shift in AI investment toward models that generate and coordinate actions, not just language. Its gameplay data gives the company a distinctive research angle, but the proposed valuation is ahead of the public evidence available about real-world robotics performance.

For AI builders and enterprise buyers, the meaningful test will be whether General Intuition can convert abundant action data and large-scale compute into dependable behavior across changing physical environments. Until the round closes and independent technical or deployment evidence emerges, the $6 billion figure should be treated as a measure of investor conviction—not a confirmed measure of product maturity.