Physical Intelligence, the two-year-old San Francisco robotics startup founded by former Google DeepMind researchers Sergey Levine and Brian Ichter alongside former Stripe executive Lachy Groom and academic co-founders, is in talks to raise about $1 billion in new funding at a valuation exceeding $11 billion — effectively doubling the $5.6 billion valuation it commanded just four months ago, according to Bloomberg reporting confirmed by TechCrunch. If the deal closes at the reported terms, Physical Intelligence will have raised roughly $2.1 billion in total capital since its founding, a number more typical of a late-stage software company than a hardware-adjacent robotics firm that currently employs around 80 people.
Founders Fund is set to participate in the round alongside Lightspeed Venture Partners, with returning backers Thrive Capital and Lux Capital also in talks. The deal is in early stages and details could shift, per Bloomberg.
The speed of the valuation jump is what made investors and observers do a double-take. In November 2025, Physical Intelligence closed a $600 million round led by CapitalG, Google's growth fund, at $5.6 billion. That round itself was large by most benchmarks. A doubling in four months signals that the investors who backed the November raise — and the new names circling this one — believe Physical Intelligence has crossed some threshold in its technology that justifies the number. The question is what that threshold actually is.
Co-founder Sergey Levine, a UC Berkeley associate professor on leave to run the company, has described Physical Intelligence's ambition plainly: /"Think of it like ChatGPT, but for robots."/ The comparison is useful shorthand for a more complicated idea — that general-purpose robot brains, like general-purpose language models, can be trained on massive datasets and then fine-tuned for specific tasks, rather than each robot requiring hand-coded behavior for every action. If that bet pays out, the winner doesn't just sell a robot — they sell the operating system that runs every robot.
Co-founder Lachy Groom, a former Stripe executive who has become one of the most active infrastructure investors in the AI space, told TechCrunch in January that the company has no timeline for commercialization. /"There's no limit to how much money we can really put to work,"/ Groom said. /"There's always more compute you can throw at the problem."/ It's a frank acknowledgment that this round, like the ones before it, is buying research velocity. The investors appear comfortable with that.
Physical Intelligence's approach is distinct from robot makers like Figure AI, valued at $39 billion and building its own proprietary model called Helix in-house. Physical Intelligence sells the model layer itself. In February 2026, the company released the Physical Intelligence Layer, which it describes as enabling partners to solve real-world robotics problems using its foundation models. The company has published research on giving Vision-Language-Action models both long-term and short-term memory — a capability relevant to manipulation tasks like laundry folding or kitchen cleanup that have historically defeated robots trained on narrow datasets. It open-sourced its π0 model in February 2025, a relatively rare move in a space where most companies treat model weights as core competitive moat.
The investor roster reflects the gravity the market is placing on this bet. Bond, Jeff Bezos, Khosla Ventures, OpenAI, Sequoia Capital, CapitalG, and Thrive Capital are all listed as backers on the company's site — a list that spans consumer tech, enterprise software, AI research, and defense-adjacent venture capital, suggesting the company is seen as infrastructure by multiple distinct buyer profiles.
The competitive landscape is where the story gets complicated. Figure AI, backed by Intel, Nvidia, Qualcomm, and Salesforce, is valued at $39 billion and building its own end-to-end AI for humanoid robots. 1X Technologies, another humanoid robotics startup, raised $100 million in a Series B round. And Google DeepMind — the former employer of co-founder Sergey Levine and former Google Research engineer Brian Ichter, per the company's founding roster — continues its own robotics research. Physical Intelligence is betting that the robot OS layer wins regardless of which hardware OEM dominates. That's a defensible position, but it's a bet on industry structure, not just technology.
What's worth watching is whether the valuation — which already prices in a world where Physical Intelligence's model runs across millions of robots — survives contact with what the company can actually ship. The funding math is straightforward if robots succeed at scale. It's considerably more speculative if the commercialization timeline stretches, or if the major hardware players build their own brains instead of licensing someone else's.
The deal, if it closes, is expected to be announced in the coming weeks.
