The boy from Perth who taught himself HTML at 10, sold his first company at 13, became Stripe's 30th hire, and is now building the robot brain that might run the physical world.
Right now, somewhere in a San Francisco lab, a robot arm is picking up a crumpled shirt and folding it. Not because a human showed it every possible shirt in every possible crumple - but because a foundation model figured it out. That model has Lachy Groom's name on it.
Physical Intelligence, the robotics AI company Groom co-founded in March 2024, is trying to do for robots what GPT did for language. Build one brain. Put it in any body. Watch it learn. By November 2025, investors valued it at $5.6 billion. By early 2026, the company was reportedly in talks for another round that would put it past $11 billion. The product - their π0 and π*0.6 models - can make espresso, fold laundry, and assemble packaging at over 90% success rates. On tasks they were never explicitly taught.
The strange part is that Groom's co-founders - Karol Hausman (former Google DeepMind), Sergey Levine (UC Berkeley), Chelsea Finn (Stanford), Brian Ichter - are some of the world's foremost robotics researchers. Groom is the one who builds things, backs things, makes things move. He is the business brain alongside some of the sharpest academic minds in AI. This is not an accident. It is a pattern.
Before Physical Intelligence, Groom ran one of Silicon Valley's most unusual investment operations: a solo GP fund - no partners, no committee, just his own conviction - that deployed capital into companies the rest of the market had not yet understood. He wrote checks of $100K to $500K at the seed stage. He backed Figma when it was valued at $94 million. Notion when the skeptics were loudest. Ramp, Lattice, Deel. And OpenAI. And Anduril. More than 200 companies in total, 11 of which became unicorns.
The Figma bet alone, at its peak $20 billion valuation, would have returned approximately 185x. When Adobe announced its $20 billion acquisition attempt, the world learned what early believers had known for years.
That run did not come from a carried interest at a major fund. It came from a kid who sold his first internet business at 13, moved to San Francisco without a college degree at 18, and somehow talked his way into Stripe during its most critical growth phase.
Perth, Australia. Age ~10. Dial-up modem.
The obsession that never stopped.
Perth, Western Australia. Dial-up internet. A grandfather patient enough to sit with a 10-year-old and explain HTML and CSS. That is how Lachy Groom's career began - not in a dorm room or a startup incubator, but at a family computer in a city that is more geographically isolated from Silicon Valley than almost anywhere on Earth.
By 13, he had built PSDtoWP.com - a service that converted Photoshop mockups into WordPress themes for freelance designers - and sold it nine months later. At the same time: PAGGStack.com, a nutritional supplements marketplace that moved product globally before he was old enough for high school exams.
At 15, he launched iPadCaseFinder.com. The timing was perfect - the iPad had just launched, the market for cases was exploding, and Groom had figured out search traffic before most adults in the industry had caught on. The site attracted 400,000 unique visitors in its opening weeks and was acquired shortly after.
Then CardNap: Australia's largest marketplace for second-hand gift cards. The concept came from watching Americans use Plastic Jungle and wondering why nothing similar existed in Australia. Groom built it. It became the market leader. He sold it before finishing high school.
At 18, Australian business magazine Anthill put him on their 30under30 list. The same year, he finished high school, skipped university, bought a one-way ticket to San Francisco, and enrolled in a Ruby on Rails bootcamp. The bootcamp startup liked his work so much they hired him directly. His projects attracted major tech press coverage. One of those articles reached the inbox of someone at Stripe.
One of Australia's most promising entrepreneurs under 30. He was 18. He had already built and sold four companies. He had not yet set foot in Silicon Valley.
The Stripe interview did not go well. The hiring committee passed. The offer did not come. But before Groom could book a flight back to Australia, an employee named Ric Burton went to bat for him - pushed back on the committee, argued the case, and secured a 30-day trial. Groom took the job. Six years later, he was running Stripe Issuing.
Stripe Issuing is the product that lets companies create, distribute, and manage physical and virtual payment cards programmatically. It is how modern fintechs - expense management tools, spend management platforms, embedded finance products - generate the cards their users carry in their wallets. Building it required understanding financial partnerships at a depth most companies never develop: direct negotiations with Visa, Mastercard, First Data.
Before that, Groom was building out Stripe's global presence. He led international expansion into Singapore, Australia, New Zealand, Hong Kong, and across Europe - each market carrying its own regulatory complexity, partner negotiations, and go-to-market puzzle. He ran the global business development and operations team. He was, in the vocabulary of Stripe internally, someone who could be handed a problem that had no obvious owner and be trusted to own it.
Stripe during those years was one of the fastest-growing companies in Silicon Valley history. Groom was employee #30. He had a front-row seat to what company-building looks like when Patrick Collison and John Collison are pushing the throttle all the way down. He absorbed that education. He used it.
In 2018, he left. Not because Stripe was faltering - it was worth tens of billions - but because he saw an opportunity to do something with the pattern recognition he had built up, and that opportunity was investing. Full-time, solo, on conviction.
Solo capitalist is a term that gets used loosely. What it actually means, at its best, is this: one person, writing checks from their own fund, making decisions without a committee, backing founders who fit a specific thesis so precisely that the check arrives before the pitch deck is finished. That is roughly how Groom operated from 2018 to 2024.
His thesis was not complicated to articulate but took real nerve to execute: back tools that developers and individual contributors fall in love with first, and let enterprises pay for it later. Bottom-up adoption. Products that spread through organizations because users pulled them in, not because a sales team pushed them down.
Figma fit the thesis. Notion fit the thesis. Ramp fit the thesis. These were not obvious bets when Groom made them. Figma had Sketch and Adobe. Notion had too many note-taking apps to count. Ramp was competing against entrenched corporate card programs. But in each case, the product-market fit signal was real, the founders were technical, and the adoption curve was starting from the bottom up.
The LPs who backed his funds - Harvard University endowment, Horsley Bridge Partners, Michael & Susan Dell Foundation, Karsh Family Foundation - are not institutions that write checks to unproven managers. They are gatekeepers who spend years evaluating track records before committing capital. Groom earned them.
His portfolio ran wider than his headline names suggest. Zepto in quick commerce. Roboflow in computer vision. Doppler in secrets management. incident.io in incident response. Modern Health in mental health. Routable in B2B payments. Over 200 companies, writing $100K to $500K checks, staying involved enough to actually help without becoming a distraction.
He also maintained a GitHub account with personal side projects - a Swift app, a forked AI chat tool, a flight price scraper. The investing career did not kill the builder instinct. It just ran parallel to it.
In March 2024, Groom co-founded Physical Intelligence alongside some of the most cited robotics researchers alive. The pitch: general-purpose AI for the physical world. Not a robot that can do one task in a controlled environment. A foundation model that can control any robot body to perform any physical task.
Their first model, π0, demonstrated it was possible. Their latest, π*0.6, went further: a robot arm, with no task-specific training, learned to fold laundry, make espresso, and assemble cardboard boxes - at over 90% success rates. The capability that matters here is the transfer - learning something new from a small amount of data, because the underlying model already understands how objects behave in the physical world.
The parallel to large language models is not an analogy. It is the actual architecture. Physical Intelligence is betting that the same recipe that cracked language - scale, data, transformers - will crack physical manipulation too. If they are right, the implications run from manufacturing to logistics to home robotics to healthcare.
Investors seem to agree. Jeff Bezos. Thrive Capital. Lux Capital. Khosla Ventures. Sequoia. And then CapitalG, Alphabet's growth fund, leading a $600M round at a $5.6 billion valuation in November 2025. By March 2026, the company was reportedly in talks for a billion more at a valuation north of $11 billion.
Groom's role here is the same role he has always played: the person who makes complex technical work land in the real world. Business, product, operations. The academic researchers understand the science. Groom understands how to build a company around it.
"My granddad actually taught me HTML when I was a kid, which got me pretty fascinated with the internet."
"Starting my business back in Australia could have been so much simpler if something like Stripe had existed."
"Sam is the most supportive, generous, inspiring person I know."
"recently i've found it hard to keep up with the pace of developments in AI. i found myself wanting a curated source of all the most noteworthy developments. so i built it!"
The thing that distinguishes Groom from a generation of people who left good tech jobs to write checks is that he never stopped building things. While running his investment funds, he built a personal AI news aggregator because he found himself overwhelmed by the pace of AI development and wanted a curated feed. He posted about it on Twitter with the same casual energy as someone describing a weekend project. Because that is what it was.
His Twitter presence - @lachygroom, 54,000+ followers - is not a polished brand play. It is a running commentary from someone who is genuinely obsessed with technology, occasionally self-deprecating, and not particularly interested in performing thought leadership. He tweets about consumer tech products. He tweets dry observations about how people use their phones. He tweets things that are funny because they are specific.
The arc from Perth dial-up to Physical Intelligence co-founder is consistent in one way: Groom has always been attracted to the moment before something becomes obvious. He built iPad accessories before the market understood the iPad. He backed Figma before the market understood browser-based design. He backed bottom-up SaaS before the industry had a word for it. And now he is betting that robots with general-purpose AI are that same moment - early, uncomfortable, and almost certainly correct.
The boy wonder tag that press applied to him early in his San Francisco career has not entirely worn off. TechCrunch described him as "fresh-faced" in a 2026 profile. At 31, having built or backed companies now worth tens of billions of dollars, he still shows up to meetings looking like the youngest person in the room. That, too, is a consistent detail.