General Intuition funding: $220M round lifts valuation to $6.2B

From reported talks to a closed round

The gameplay-trained AI lab General Intuition has gone from reported talks to a closed round in about five weeks, and the result tells you a lot about where venture money is going in late 2026. On August 24, reports said the New York startup was negotiating a raise at a $6 billion pre-money valuation, with Valor Equity Partners, Point72 Ventures and Alexis Ohanian's Seven Seven Six joining as new investors.12 At the end of September, the company said it had closed $220 million at a $6.2 billion valuation. Atreides was added to the investor list, alongside Valor, Seven Seven Six, Point72 and existing backers Khosla Ventures and General Catalyst.1214

The valuation is the headline, but the pace stands out more. General Intuition announced a $133.7 million seed round in October 2025 and a $320 million Series A at a $2.3 billion valuation on June 25, 2026.11 The August talks came out exactly 60 days after that Series A.8 Adding the new round, disclosed funding is now about $673.7 million.1418 In under a year, a spinout from a gaming clip-sharing app has become one of the most richly valued companies in physical AI.

The bet: gamers' button presses as robot training data

The investment case depends on one dataset. CEO Pim de Witte spun General Intuition out of Medal, the clip-sharing platform he founded. The company uses Medal's hundreds of millions of hours of gameplay, paired with "action labels" that record which buttons a player pressed and when, as its starting training data.1 Medal has about 17 million monthly active users, according to the company.11 General Intuition also says the platform is on track for three billion uploaded videos a year. That figure, and the company's claims about how its data scale compares with rivals, come from the company itself.14

The argument is that video alone is not enough. Pairing what a player sees with what the player did lets a model link perception to action and to the result that followed.14 Vinod Khosla, whose firm led both earlier rounds, has said these labels will be central to what he calls the "emergence of intuition": a model's ability to handle tasks it was never explicitly trained on.1 Khosla has also said that buying the company now would amount to a data acquisition, which he considers uninteresting. That fits earlier reports that OpenAI once looked at acquiring Medal for about $500 million.1619

The framing has shifted since the August reports. Most coverage of the August talks called General Intuition a world-model company pushing into robotics.46 By the time the round closed, the company was stressing "action models": systems that, given an environment, a body and a goal, decide what to do. World models, which generate the environment itself, are used only as internal training tools.13 That matches what de Witte showed reporters in June, when he described the world model as a training "gym" and the agent model as the product. This doesn't change the business, but it does sharpen the pitch. General Intuition wants to sell the decision-making layer, not the simulation.

Where the coverage agrees and where it splits

Most outlets agree on the main facts: the investors, the Medal origin, the CoreWeave compute partnership, and the plan to spend on compute, hiring and robotic embodiments.356 Several details vary, though, and some of the differences matter.

First, the valuation basis. August reports consistently described $6 billion as pre-money.110 The September announcement gave $6.2 billion without saying whether that was pre- or post-money.14 If you assume the round priced at the reported $6 billion pre-money, adding $220 million gives roughly $6.2 billion post-money, so the two figures look like the same deal described differently. Read that way, the valuation didn't rise between the talks and the close. Comparisons with June also differ: one outlet called the jump 2.3x, another roughly 2.6x, and others "nearly triple."13210 Which multiple you get depends on which valuation bases you compare.

Second, who led. Some outlets said the round was led by Valor, Point72 and Seven Seven Six.46 Others described them only as new participants, and one investor database marks the lead designations as unverified.87 Naming also drifted: several reports said "Valor Ventures" rather than Valor Equity Partners.24

Third, basic chronology. One summary put the Medal spinout in October 2024, while most reporting puts it in October 2025.176 A media-comparison analysis also noted that some coverage mixed up General Intuition with Generalist, a different robotics startup founded by former Google DeepMind researchers, at a different valuation.3 A crowded field of similarly named physical-AI companies makes those errors more likely.

One detail stands out for the venture angle. Reports said Valor, best known for backing SpaceX, would be making its first AI-lab investment since SpaceX.35 Valor is known for funding capital-heavy companies that operate in the physical world.9 Its interest suggests General Intuition is being valued less like a software company and more like an infrastructure project.

A crowded field of physical-AI bets

General Intuition is raising into what may be the busiest corner of AI funding outside the large language-model labs. World Labs raised $1 billion in February for spatial intelligence, and Skild AI announced a $1.4 billion Series C in January at a valuation above $14 billion.8 Decart raised $300 million in May and Odyssey $310 million in June, part of a wider wave of world-model funding.19 One report says AMD has agreed to acquire World Labs for $8.2 billion, with founder Fei-Fei Li joining AMD. Wider confirmation of that deal is still limited.13 Big tech is moving in too: Alibaba released a suite of robot models in June that includes navigation, world-model and manipulation systems.11

These companies are taking different routes to the same goal. Physical Intelligence has built on teleoperation data from real robots, while Skild leans on simulation.9 General Intuition's route is different: a huge stream of human interaction data it already owns, which reduces the need to collect every training example on physical hardware.8 One investor-oriented analysis argues that robotics foundation models are attracting more AI venture money partly because the language-model tier is now too expensive for most funds to own meaningful stakes.9 That makes sense. At a $6.2 billion valuation, General Intuition still offers growth-stage investors a way into AI's next platform before prices climb further.

The gap the valuation has to close

The most careful coverage keeps returning to one point: the transfer from games to robots hasn't been proven at commercial scale. In June's demo, the same model ran a game agent and a quadruped robot in the company's office. The robot needed just eight minutes of real-world data for fine-tuning, but it still clipped chair legs and bumped into bins.8 Reliable physical-world performance at scale is still unsolved across this whole class of systems.14 One venture analyst argued the key test is a head-to-head comparison: a game-pretrained policy against a model trained on teleoperation data, using the same benchmark and fine-tuning budget. Without that, the data advantage is a story rather than a moat.915

The commercial picture is also early. General Intuition is testing a commercial version with a small group of customers in robotics, simulation and entertainment. It opened a waitlist when it announced the funding and says the money will mainly go to hiring researchers.12 The company plans to grow its team in New York and Europe.13 It is positioning itself as a hardware-agnostic intelligence layer rather than a robot maker, so each customer could also supply new interaction data from different machines.8 It hasn't publicly named its partners.8

The takeaway

This round says less about General Intuition's technology than about what investors will pay for proprietary data in 2026. With few clean, action-labeled datasets for physical AI, investors are treating Medal's archive as a scarce resource and pricing it ahead of proof. The Series A paid for compute; this round adds researchers and brings in growth-stage investors, including a SpaceX backer making its first AI-lab bet since SpaceX.16125

The bet is easy to understand, and the valuation is not unreasonable given what comparable physical-AI companies have raised. But General Intuition now carries a valuation that assumes gameplay data will transfer to real machines, not just that it might. Its next round will probably depend on whether it can publish or show results where its game-trained models clearly beat robot-native training on real tasks. If it can, $6.2 billion could look cheap. If it can't, Medal's archive could turn out to be the kind of data acquisition Khosla called uninteresting.