Robotics AI Startup Physical Intelligence Seeks Billion at Billion Valuation

Physical Intelligence, the San Francisco-based robotics startup developing general-purpose AI models for robots, is in early talks to raise billion in new funding — a round that would nearly double its valuation to over billion.

The company, often described as "ChatGPT for robots," was founded in 2024 by AI academics and former Google DeepMind researchers. Its mission is to build AI systems that enable robots to handle a wide range of tasks in logistics, manufacturing, and healthcare, without requiring specialized programming for each individual job.

From .4 Billion to Billion in Two Years

Physical Intelligence has grown at remarkable speed. In 2024, it raised million at a .4 billion valuation, followed by a million round that valued the company at .6 billion. The new round, reported on March 28, 2026, would push the valuation to over billion, more than doubling the previous figure.

Investors reportedly in discussions for the current round include Peter Thiel's Founders Fund, Lightspeed Venture Partners, Thrive Capital, and Lux Capital. Previous backers include Sequoia Capital, Khosla Ventures, CapitalG (Alphabet's growth fund), Jeff Bezos, and OpenAI.

What Does Physical Intelligence Actually Build?

The company is focused on building "foundation models" for robotics, similar to large language models from Anthropic and OpenAI, but adapted for physical machines. These models aim to give robots the ability to quickly learn and adapt to new tasks without the traditionally lengthy custom programming required for each use case.

The market potential is enormous, but so are the technical challenges. The robotics industry has long struggled with the transition from controlled lab environments to the chaotic reality of warehouses and factories.

The CIO Perspective

For technology leaders in manufacturing and logistics, this is a signal worth tracking closely. When large capital starts flowing into general-purpose AI models for robots, we approach a threshold where automation becomes dramatically more accessible. No more months of custom integration per robot task.

The next two to three years will reveal whether the promises of generalizable robot AI hold up in production.