Physical Intelligence, Stripe veteran Lachy Groom's latest bet, is building Silicon Valley's buzziest robot brains - TechCrunch
Inside, the space is a giant concrete box made slightly less austere by a haphazard sprawl of long blonde-wood tables. Some are clearly meant for lunch, dotted with Girl Scout cookie boxes, jars of Vegemite (someone here is Australian), and small wire baske...
Key takeaways
The most recent coverage shows that humanoid robots are moving from prototype showcases toward modest commercial roll‑outs, but significant technical and economic gaps remain. A market report released in late January counted about 16,000 installations worldwide in 2025, with Shanghai‑based AgiBot leading the field at roughly 31 % of the market, followed by Unitree (27 %) and UBTech (just over 5 %). Tesla’s Optimus line entered the top five with close to a 5 % share, while the top five suppliers together accounted for roughly 73 % of all deployments. Companies are experimenting with lower‑cost, interaction‑focused platforms and robot‑as‑a‑service models, especially in China, where leasing humanoids for retail, performances and promotional events is gaining traction. At the same time, industry insiders caution that most current robots still struggle with safety, battery life, dexterity and cost‑effectiveness outside controlled settings, a point underscored by the McKinsey “Humanoid robots: Crossing the chasm” analysis and by live mishaps such as XPeng’s IRON robot falling face‑first during its debut. Nonetheless, high‑profile demos have highlighted progress: Boston Dynamics’ Atlas performed tasks at Hyundai’s Savannah plant, while the British startup Humanoid unveiled its HMND 01 biped at CES 2026, claiming 25,000 pre‑orders and successful pilots with Fortune 500 firms. In China, PNDbotics released a video of its Adam‑U Ultra dancing with precise, human‑like motion, and the sector continues to attract substantial investment, with $4.6 billion poured into humanoid development in 2025 alone.
Inside, the space is a giant concrete box made slightly less austere by a haphazard sprawl of long blonde-wood tables. Some are clearly meant for lunch, dotted with Girl Scout cookie boxes, jars of Vegemite (someone here is Australian), and small wire baskets stuffed with one too many condiments. The rest of the tables tell a different story entirely. Many more of them are laden with monitors, spare robotics parts, tangles of black wire, and fully assembled robotic arms in various states of attempting to master the mundane. Quan Vuong, another cofounder who came from Google DeepMind, explains that the strategy revolves around cross-embodiment learning and diverse data sources. If someone builds a new hardware platform tomorrow, they won’t need to start data collection from scratch – they can transfer all the knowledge the model already has. “The marginal cost of onboarding autonomy to a new robot platform, whatever that platform might be, it’s just a lot lower,” he says. During my visit, one arm is folding a pair of black pants, or trying to. It’s not going well. Another is attempting to turn a shirt inside out with the kind of determination that suggests it will eventually succeed, just not today. A third – this one seems to have found its calling – is quickly peeling a zucchini, after which it is supposed to deposit the shavings into a separate container. The shavings are going well, at least.