Physical Intelligence is in talks to raise at least $1 billion at a valuation of roughly $20 billion, Bloomberg reported on October 27, 2026. The proposed financing would give the robotics AI startup substantial new capital as it expands outside access to models intended to control different kinds of robots.
Existing investors Avenir and Lux Capital are discussing leading the round, according to Bloomberg’s anonymous sources. The talks are not a completed financing: the amount, valuation and participants could change, or the transaction could fail to close. Physical Intelligence, Avenir and Lux Capital did not respond to Bloomberg’s requests for comment, according to Quartz’s account of the report.
The timing is worth noting. Physical Intelligence reportedly began offering outside developers beta access to its newest robotics models earlier in October. Investors are considering a much higher valuation while developers gain a chance to evaluate and adapt the technology. Those are two different milestones, and neither yet provides a public measure of the company’s commercial adoption.
What a $20 Billion Valuation Would Mean
Physical Intelligence previously secured a valuation of about $11.5 billion earlier in 2026, according to Quartz, citing Bloomberg. Moving to $20 billion would represent an increase of roughly $8.5 billion, or 74%, based on those reported figures.
That is a substantial increase within the same year. It also gives more precision to the description that the valuation would “nearly double”: the proposed price is about 1.74 times the previous one.
The company reportedly has no revenue. That makes the proposed valuation a bet on future adoption of its robotics software, with investors assigning value to the possibility that its models could become useful across many applications and hardware platforms. It is not a valuation that can be assessed against an existing sales base.
My view is that the combination matters more than either headline number alone. A proposed billion-dollar round suggests investors see an opportunity worth financing at scale. A roughly $20 billion valuation for a pre-revenue company means much of that opportunity still has to become a business.
The available reporting does not specify whether the proposed valuation is before or after the new investment. That distinction affects ownership and dilution, so dividing the proposed investment by the headline valuation would not reliably establish the stake investors would receive.
Beta Access Opens a New Evaluation Stage
Physical Intelligence began letting outside developers use and tailor its latest robotics models through a beta program earlier in October, according to Quartz. Investing.com’s summary describes that access as allowing developers to fine-tune the models and says the rollout helped spur investor interest.
Fine-tuning generally means adapting an existing model through additional training for a particular task or setting. For a robotics developer, the attraction is straightforward: starting with a reusable model could reduce the work needed to build the intelligence behind each application from scratch.
Physical Intelligence says its software works across hardware from different manufacturers and supports many kinds of robot motion, according to the coverage. That is the central promise of a general-purpose robotics foundation model: a shared software base that can be adapted to multiple robots and jobs.
Broader access is a meaningful development, but beta availability is not the same as a production-ready deployment. The supplied reporting does not detail supported robot configurations, access pricing, licensing terms, performance guarantees or safety requirements. It also does not establish that the models are open source or generally available without restrictions.
That gap is particularly important in robotics. An application needs a model that performs the task, but it also needs suitable hardware, integration with surrounding systems, ways to handle failures and a workable operating process. A model provider and an application developer can therefore make progress on different schedules.
I would read the beta as an opportunity for more concrete evaluation. Developers can begin examining whether a shared model fits their particular workflow. The existence of that opportunity is useful; the results of those evaluations will be more informative than the access announcement alone.
The Competitive Comparison Has Limits
The reported round places Physical Intelligence in a field that includes specialist robotics AI companies and much larger technology businesses. The coverage identifies OpenAI, Tesla and Skild AI as competitors, although it provides no matched technical benchmarks showing how their systems compare.
Skild AI offers a useful financial comparison because the reporting includes both a valuation milestone and a commercial metric.
| Company | Reported financing milestone | Reported commercial position |
|---|---|---|
| Physical Intelligence | Previously valued at about $11.5 billion in 2026; now discussing at least $1 billion at roughly $20 billion | Reportedly has no revenue; outside model access is in beta |
| Skild AI | Reached a $14 billion valuation in a January 2026 financing | Announced in September that annual recurring revenue had passed $100 million |
These figures come from Quartz’s and Investing.com’s accounts of Bloomberg’s reporting. They are not independent confirmations of the financing terms or revenue figures. The available evidence is one reporting chain, and Skild’s revenue milestone is attributed to the company.
The comparison nevertheless highlights an important distinction. Physical Intelligence’s proposed valuation would exceed Skild’s reported January valuation, despite Physical Intelligence reportedly having no revenue. The figures come from different dates and financing circumstances, so they cannot establish which company is worth more on equivalent terms.
Annual recurring revenue also needs careful interpretation. It is an annualized measure of recurring business, not necessarily revenue recognized over a completed year, cash collected or profit. Skild’s reported milestone indicates a different stage of commercialization, but it does not establish superior robotics performance or better economics for a particular customer.
The useful conclusion is narrower: investor pricing, technical capability and commercial traction are separate measures. A large valuation can coexist with an early commercial position. Revenue can show that customers are paying without proving that a system fits every deployment.
What Prospective Users Should Evaluate
Enterprise buyers considering Physical Intelligence’s beta have a more practical decision than whether the startup deserves a $20 billion valuation. They need to establish whether its models can support their own tasks at an acceptable cost and level of risk.
Frequently Asked Questions
3 questions
1Has Physical Intelligence completed the $1 billion round?
No completed round is established by the available reporting. Bloomberg reported negotiations for at least $1 billion at a valuation of roughly $20 billion, with terms still subject to change.
