Less than a year after closing one of the largest funding rounds in robotics history, Physical Intelligence is back at the table. The San Francisco-based startup is reportedly in talks to raise another $1 billion, a move that would value the company at approximately $4.5 billion and cement its position as the most aggressively capitalized player in the race to build general-purpose robot brains.
The round, first reported by TechCrunch, would match the size of the company’s previous fundraise completed in late 2025. That earlier round, led by Jeff Bezos and Thrive Capital, valued Physical Intelligence at roughly $2.4 billion — a staggering figure for a company that was barely a year old at the time and had yet to ship a commercial product.
Now the valuation could nearly double.
The speed of Physical Intelligence’s ascent tells a broader story about where Silicon Valley’s biggest checks are flowing. After years of pouring capital into large language models and chatbots, investors are pivoting hard toward companies that promise to bring AI out of the screen and into the physical world. Robotics, long considered too capital-intensive and too slow for venture-style returns, has become the hottest sector in tech funding.
The Bet Behind the Billions
Physical Intelligence, often referred to as Pi, was founded in 2024 by a team of researchers with deep roots in robotics and machine learning. The founding group includes Karol Hausman, formerly of Google DeepMind, and Sergey Levine from UC Berkeley, both widely regarded as leading figures in robot learning research. Their thesis is deceptively simple: build a single AI model — a foundation model — that can control many different types of robots performing many different tasks.
This is harder than it sounds. Traditional robotics programming requires painstaking, task-specific engineering. Pick up a cup? That’s one program. Fold a shirt? An entirely different one. Pi’s approach borrows from the playbook that made ChatGPT possible — train a single large model on vast amounts of data and let it generalize across situations it hasn’t explicitly been programmed to handle.
The company’s flagship effort is what it calls π₀ (pi-zero), a vision-language-action model designed to translate high-level instructions into precise physical movements. In demonstrations, the model has shown the ability to fold laundry, pack items into boxes, and manipulate objects with a dexterity that has impressed even skeptics in the robotics community.
But demonstrations aren’t products. And that’s the tension at the heart of Pi’s fundraising blitz.
The company has not yet announced significant commercial deployments or revenue figures. It’s raising capital at a pace that assumes the technology will work at scale, in messy real-world environments, on hardware it doesn’t control. That’s a massive leap of faith — one that investors seem eager to make.
According to TechCrunch, the new round’s investor lineup has not been finalized, though the publication noted that Physical Intelligence’s existing backers include some of the most prominent names in venture capital: Bezos Expeditions, Thrive Capital, Lux Capital, and Khosla Ventures, among others. Whether all or some will participate in this latest raise remains unclear.
The $4.5 billion valuation target is aggressive by any standard. For context, Boston Dynamics — which has been building and iterating on advanced robots for more than three decades — was acquired by Hyundai in 2020 for approximately $1.1 billion. Pi would be valued at more than four times that figure with a fraction of the operational history.
A Market Suddenly Crowded With Capital
Physical Intelligence isn’t raising in a vacuum. The broader robotics and embodied AI sector has seen an extraordinary influx of funding over the past 18 months. Figure AI, another humanoid robotics startup, raised $675 million in early 2024 at a $2.6 billion valuation, with backing from Microsoft, OpenAI, and Jeff Bezos. Norwegian humanoid startup 1X Technologies secured $100 million in a Series B. Chinese robotics firms, including Unitree and Agibot, have attracted significant state-backed and private investment.
The common thread: investors believe that foundation models for robotics could do for physical automation what large language models did for digital tasks. The potential market is enormous — manufacturing, logistics, elder care, agriculture, household assistance. Goldman Sachs has projected that the humanoid robot market alone could reach $38 billion by 2035.
And yet the technical challenges remain formidable. Robots operating in unstructured environments face problems that chatbots never encounter. Physics doesn’t forgive hallucinations. A language model that produces a slightly wrong answer is an inconvenience; a robot arm that applies slightly wrong force can destroy whatever it’s handling — or injure a person standing nearby.
Pi’s approach of building a generalist model rather than a task-specific one is intellectually compelling but commercially unproven at scale. The company must demonstrate that π₀ can transfer from controlled lab settings to factory floors, warehouses, and eventually homes. That transition has historically been where robotics startups stumble.
So why are investors so willing to write billion-dollar checks?
Part of the answer is FOMO — the fear of missing the next platform shift. Early investors in OpenAI saw returns that reshaped their portfolios. The thinking goes that if embodied AI follows a similar trajectory, being early in the right company could generate outsized returns. Part of the answer is also structural: the cost of training large models is enormous, and companies that can’t raise sufficient capital simply can’t compete. This creates a winner-take-most dynamic that rewards aggressive fundraising.
There’s also genuine technical momentum. Research from Pi and competitors has shown rapid improvement in robot dexterity and task generalization over the past two years. The models are getting better fast. Whether they’re getting better fast enough to justify current valuations is the billion-dollar question — literally.
Pi’s team has been strategic about positioning. Rather than building its own humanoid hardware — an expensive and time-consuming endeavor — the company has focused on the software layer, designing models that can be deployed across different robot form factors from various manufacturers. This asset-light approach is more attractive to venture investors than the vertically integrated model pursued by companies like Figure AI or Tesla’s Optimus program, which must bear the costs of both hardware development and AI research simultaneously.
The strategy carries risks. By depending on third-party hardware, Pi cedes control over a critical part of the user experience. If the robots running its software are unreliable or poorly designed, Pi’s models look bad regardless of their sophistication. It’s a familiar tension in technology — the platform play is higher margin but lower control.
What Comes Next
If the $1 billion round closes as reported, Physical Intelligence will have raised approximately $2 billion in total funding within roughly two years of its founding. That’s an extraordinary war chest. It’s also an extraordinary amount of expectations to carry.
The pressure to show commercial traction will intensify. Investors at these valuations aren’t buying a research lab; they’re buying a future market leader. Pi will need to announce partnerships, pilot programs, or early revenue streams that validate its technology outside the lab. The company has reportedly been in discussions with logistics and manufacturing firms, though details remain scarce.
Competition is accelerating from multiple directions. Google DeepMind continues to invest heavily in robotics research. OpenAI has signaled interest in embodied AI. And a growing cohort of well-funded startups — Skild AI, Covariant (now part of Amazon), and others — are pursuing overlapping visions of general-purpose robot intelligence.
Then there’s the China factor. Chinese companies are moving fast in humanoid robotics, often with substantial government support and access to manufacturing infrastructure that U.S. startups can’t match. The geopolitical dimension adds urgency to domestic funding efforts but also raises questions about market access and technology transfer.
For now, Physical Intelligence occupies an enviable position: well-funded, well-staffed, and working on a problem that the market has decided is worth solving at almost any price. The next 18 months will reveal whether that confidence is justified or whether the company becomes another cautionary tale of hype outrunning reality.
The robotics industry has seen this movie before. What’s different this time is the scale of the bets — and the caliber of the people making them.
WebProNews is an iEntry Publication