$320m raise values robotics AI startup General Intuition at $2.3bn

General Intuition, a startup developing foundational artificial intelligence for robotics, raised $320 million last month at a $2.3 billion valuation. The funding round was led by Vinod Khosla. The company is betting that the robotics industry is about to undergo a structural shift away from specialized, data-heavy models toward general-purpose systems.

The startup's thesis mirrors the evolution of text-based AI. Before large language models, companies built task-specific natural language processing tools from scratch. Today, most rely on general models and fine-tune them. De Witte believes embodied AI will follow the same path, arguing the industry should focus on datasets that produce foundation models capable of transferring intuition across environments.

Currently, robotics firms spend vast amounts of time and money gathering real-world data to train machines for specific physical spaces. General Intuition bypassed this bottleneck by training its model on millions of hours of video game data. By analyzing what buttons a human pushed and when, the system developed a human-like intuition for spatial-temporal reasoning, which de Witte and Khosla view as the missing link for physical AI.

The company validated its approach by using the model to power a four-legged robot. After fine-tuning it with just eight minutes of real-world data, the machine navigated a dynamic office using only a single front camera. “The fact that [the robot] was actually able to zero-shot on just the front camera, with no other sensors, in the office with dynamic objects being introduced and people walking by was a very big surprise to us. I think it’s a sign of what’s to come,” de Witte says.

For Europe’s vast industrial and logistics sectors, this development carries significant economic implications. If a foundation model can handle complex, unpredictable physical environments with mere minutes of training, the cost and time required to deploy industrial automation could plummet. It threatens to render redundant the millions of hours of specialized data collection currently underway across the robotics industry.

“The generalization of the model itself is the product. The fact that it has a base level of reasoning about space and time is going to be the reason why people stop collecting hundreds of thousands or millions of hours of real-world data. Because the reality is, you only need a few minutes,” de Witte said.

General Intuition does not intend to build its own hardware. Instead, its business model relies on becoming the foundational layer of physical AI, supplying the base model for other robotics companies to build upon. “We’re not gonna build a self-driving car company. We’re gonna make it 10 times easier for the next person to build a self-driving car company,” de Witte said.