From the street, the only clue I found of Physical Intelligence’s headquarters in San Francisco is a pi symbol that’s a slightly different color than the rest of the door. When I walk in, I am immediately confronted with activity. There is no front desk, no logo flashing in fluorescent lights.
Inside, the space is a giant concrete box made slightly less austere by a random row of long blond wooden tables. Some are clearly meant for lunch, dotted with boxes of Girl Scout cookies, jars of Vegemite (someone here is Australian) and little wire baskets filled with one too many condiments. The rest of the paintings tell a completely different story. Many more of them are loaded with displays, robotics parts, tangles of black wire, and fully assembled robotic arms in various states of trying to dominate the mundane.
During my visit, one hand is folding a pair of black pants, or trying to. It’s not going well. Another tries to turn a shirt inside out with the kind of determination that suggests he will eventually succeed, just not today. A third – this one seems to have found his appeal – quickly peels a pumpkin, after which he is supposed to deposit the shavings in a separate container. The wood chips are doing well, at least.
“Think of it like ChatGPT, but for robots,” Sergey Levine tells me, gesturing to the motorized ballet that unfolds across the room. Levine, an associate professor at UC Berkeley and one of the co-founders of Physical Intelligence, has the friendly, bespectacled demeanor of someone who has spent a lot of time explaining complex concepts to people who don’t immediately understand them.
What I’m watching, he explains, is the testing phase of a continuous loop: data is collected at robot stations here and in other locations—warehouses, homes, where the team can set up shop—and the data trains general-purpose robotic models. When researchers train a new model, it goes back to stations like these for evaluation. The pants-envelope is someone’s experiment. So is the shirt-turner. The zucchini peeler can test whether the model can generalize to different vegetables, learning the fundamental peeling motions well enough to handle an apple or potato it has never encountered.
THE company also operates a test kitchen in that building and elsewhere using off-the-shelf equipment to expose the robots to different environments and challenges. There’s a sophisticated espresso machine nearby, and I assume it’s for the staff until Levine clarifies that no, it’s there for the robots to learn. Any frothy latte is a given, not a perk for the dozens of engineers on stage who are mostly staring at their computers or hovering over their mechanized experiments.
The material itself is deliberately unlikely. These guns sell for about $3,500, and that’s with what Levine describes as “a huge fine” from the seller. If they built them in-house, the hardware cost would drop below $1,000. A few years ago, he says, a roboticist would have been shocked that these things could do anything. But that’s the thing – good intelligence makes up for bad hardware.
Techcrunch event
Boston, MA
|
June 23, 2026
As Levine excuses himself, I’m approached by Lachy Groom, moving through the space with the deliberateness of someone who has half a dozen things going on at once. At 31, Groom still has the fresh quality of a Silicon Valley boy wonder, a moniker he earned early, having sold his first company nine months after starting it aged 13 in his native Australia (that explains Vegemite).
When I first approached him earlier, as he welcomed a small group of sweatshirt-wearing visitors to the building, his response to my request for time with him was immediate: “Absolutely not, I have meetings.” Now he has 10 minutes, maybe.
Groom found what he was looking for when he started following academic work from the labs of Levine and Chelsea Finn, a former Berkeley PhD student of Levine’s who now has her own lab at Stanford focused on robotic learning. Their names continued to appear in anything interesting happening in robotics. When he heard rumors that they might be up to something, he tracked down Karol Hausman, a Google DeepMind researcher who also taught at Stanford and Groom had learned was involved. “It was just one of those meetings where you walk away and it’s like, this is it.”
Groom never intended to be a full-time investor, he tells me, although some might wonder why not considering his track record. After leaving Stripe, where he was an early employee, he spent about five years as an angel investor, making early bets on companies like Figma, Notion, Ramp and Lattice while looking for the right company to start or join. His first investment in robotics, Standard Bots, came in 2021 and brought him back to a field he loved as a kid building Lego Mindstorms. As he jokes, he was “on vacation much more as an investor.” But investing was just a way to stay active and meet people, not the end game. “I’ve been looking for five years for the company to launch after Stripe,” he says. “Good ideas at a good time with a good team — [that’s] extremely rare. It’s all execution, but you can execute like hell on a bad idea, and it’s still a bad idea.”
The two-year-old company has now grown over $1 billionand when I ask about his treadmill, he’s quick to clarify that he doesn’t actually burn that much. Most of its costs go into computing. A moment later, he recognizes that with the right terms, with the right partners, he will raise more. “There’s no limit to how much money we can really put into the job,” he says. “There’s always more calculation you can throw at the problem.”
What makes this arrangement particularly unusual is what Groom doesn’t give his backers: a timeline for turning Natural Intelligence into a money-making endeavor. “I’m not giving investors answers about commercialization,” he says of backers that include Khosla Ventures, Sequoia Capital and Thrive Capital, which have valued the company at $5.6 billion. “That’s kind of weird, that people put up with that.” They tolerate it, but they do, and they may not always, so the company must be well capitalized now.
So what is strategy if not commercialization? Quan Vuong, another co-founder who came from Google DeepMind, explains that it revolves around cross-embedded learning and different data sources. If someone builds a new hardware platform tomorrow, they won’t have to start collecting data from scratch — they can transfer all the knowledge the model already has. “The marginal cost of integrating autonomy into a new robot platform, whatever that platform is, is just much lower,” he says.
The company is already working with a small number of companies in different industries—logistics, grocery, a chocolate factory across the street—to test whether their systems are good enough for real-world automation. Vuong claims that in some cases, it already is. With their “any platform, any task” approach, the surface for success is large enough to start auditing automation-ready tasks today.
Natural Intelligence is not alone in pursuing this vision. The race to create general-purpose robotic intelligence—the foundation upon which more specialized applications can be built, such as the LLM models that wowed the world three years ago—is heating up. Pittsburgh-based Skild AI, founded in 2023, just this month raised $1.4 billion at a $14 billion valuation, and is taking a significantly different approach. While Natural Intelligence remains focused on pure research, Skild AI has already commercially developed the “pantosomal” Skild Brain, saying it brought in $30 million in revenue in a few months last year from security, warehousing and manufacturing.
Skild even has public shots at competitors, arguing on her blog that most “robotics foundation models” are just vision language models “in disguise” that lack “true physical common sense” because they rely heavily on Internet-scale pre-training rather than physics-based simulation and real robotics data.
It’s a pretty stark philosophical divide. Skild AI is betting that commercial development creates a flywheel of data that improves the model with each real-world use case. Natural Intelligence is betting that resisting the pull of short-term commercialization will allow it to produce superior general intelligence. Who is “right” will take years to sort out.
Meanwhile, Natural Intelligence operates with what Groom describes as unusual clarity. “It’s such a clean company. A researcher has a need, we go and collect data to support that need—or new material or whatever—and then we do it. It’s not externally driven.” The company had a 5 to 10 year road map of what it believed would be possible. By month 18, they were over it, he says.
The company has about 80 employees and plans to grow, though Groom hopes “as slowly as possible.” What is most difficult, he says, is the material. “Hardware is very difficult. Everything we do is much more difficult than a software company.” Hardware failures. It arrives late, delaying testing. Security issues complicate everything.
As Groom jumps off to rush off to his next engagement, I’m left to watch the robots continue their practice. The pants are still not quite folded. The jersey remains stubbornly right-outside. The zucchini chips pile up nicely.
There are obvious questions, including my own, about whether anyone really wants a robot in their kitchen peeling vegetables, about safety, about whether dogs go crazy with mechanical invaders in their homes, about whether all the time and money invested here is solving big enough problems or creating new ones. Meanwhile, outsiders question the company’s progress, whether its vision is achievable and whether it makes sense to bet on general intelligence rather than specific applications.
If the Groom has doubts, he doesn’t show it. He works with people who have been working on this problem for decades and who believe the time is finally right, that’s all he needs to know.
After all, Silicon Valley supports people like Groom and gives them a lot of rope from the start of the industry, knowing that there’s a good chance that even without a clear path to commercialization, even without a timeline, even without certainty of what the market will be like when they get there, they’ll figure it out. It doesn’t always work out. But when he does, he tends to make excuses many times that he didn’t.
