Physical Intelligence, a robotics AI start-up founded in 2024, is in talks to raise at least $1 billion at a valuation of about $20 billion, Bloomberg reported on October 27, 2026, citing people familiar with the matter. Existing investors Avenir and Lux Capital are in talks to lead the round, which would nearly double the company’s $11.5 billion valuation from earlier this year, even though the company has no revenue yet, and neither the company nor the reported lead investors responded to requests for comment.
A Valuation Without Revenue
The most striking fact in the report is not the size of the round but the absence behind it. Physical Intelligence, founded by former Google DeepMind researchers and academics from Stanford University and the University of California, Berkeley, builds general-purpose AI software designed to run on robots made by other companies. Its backers include Alphabet’s CapitalG, Jeff Bezos, and Thrive Capital, and earlier this month it opened its latest robot model to outside developers for fine-tuning, though the system is still in testing.
A $20 billion valuation for a company with no revenue is a bet on a future market, not a reward for a present one. Investors are paying, in effect, for a seat in what they believe will be the defining layer of physical AI: the software brain that lets a robot in a warehouse, a home, or a factory understand a task described in language and vision, and turn it into actions. If that layer becomes standard, owning a share of it could be worth far more than $20 billion. If it does not, or if a rival’s approach wins, the same stake could be worth very little.
Why Investors Are Paying Now
The timing of the talks follows a pattern across robotics funding. Rival Skild AI reached a $14 billion valuation in January and has since passed $100 million in annual recurring revenue, while Bedrock Robotics raised $600 million at an approximately $4.1 billion valuation. Physical Intelligence’s reported round would value it above both, despite having no revenue to show, which suggests its investors believe its general-purpose approach, one model for many kinds of robot hardware, is worth a premium over more specialised rivals.
Opening its model to outside developers in beta earlier this month may have helped make that case. A model that developers can try, adapt, and build on is easier to value than one locked inside a laboratory, because it creates evidence, however early, that other people want to use it. Recent job postings saying employees will train robots to complete everyday household tasks point in the same direction: the company is pitching household work, not just warehouse work, as its long-term market.
What Is Still Only Talk
It is important to be precise about the status of this deal. Bloomberg’s report describes talks, not a closed round, and the valuation, the size of the round, and the identity of the lead investors could all change before any agreement is signed, or the talks could fall apart. The company has no revenue, so there is no financial performance to anchor the $20 billion figure, only investor appetite and comparable deals. The report also does not say how much of the round would be new money, what the money would be spent on, or when it would close.
That uncertainty is normal for a private funding round reported before it closes, but it matters for readers. A $20 billion valuation in talks is a signal of investor confidence, not a market price, and it should not be read as proof that Physical Intelligence’s robots work at commercial scale. The company’s model is still in testing, and the hard problems of robotics, safety, reliability, and cost in the real world, remain unsolved for the whole field.
Why It Matters
This matters because it shows how much capital is now chasing the idea that AI’s next frontier is physical, not just digital. The same logic is driving large bets elsewhere, from General Compute’s $100 million raise to build a neocloud on rival chips to Disruptive’s $10 billion fund for AI and defence companies, where investors are funding infrastructure and platforms before the revenue arrives. If Physical Intelligence closes at anything near $20 billion, it will set a new benchmark for robotics software, and it will raise the bar for every rival trying to raise money on a smaller valuation.
It also raises a question the whole AI industry is now facing: how long can valuations run ahead of revenue before investors demand proof? Physical Intelligence’s backers are betting that the answer, for robotics, is a long time yet. The next test will be whether outside developers, given beta access to its model, build anything that ordinary people actually use.
Background
Physical Intelligence was founded in 2024 by former Google DeepMind researchers and academics from Stanford University and the University of California, Berkeley. It develops AI software that uses vision and language inputs to generate robot actions, designed to work across hardware from different manufacturers. Skild AI, a rival in general-purpose robotics, was valued at $14 billion in January 2026 and later reported more than $100 million in annual recurring revenue.
Source: Bloomberg report via Tech in Asia, Bloomberg report via PANews
0 Comments