This week, Unitree Robotics launched what it calls the world's first humanoid robot motion App Store, inviting users to download skills for their machines the way you would install an app on a phone. Independently, SAP and its partner Cyberwave announced that autonomous robots are now handling box folding, packaging, and shipping fulfillment in a live logistics warehouse in St. Leon-Rot, Germany. Two European and Chinese robotics startups, Sereact and Vbot, pulled in a combined $183 million in new funding. If you read only the press releases, it looks like the humanoid robot revolution is arriving on schedule, or perhaps even early.
James Vincent is not so sure. The journalist, who wrote a cover story on humanoid robots for Harper's Magazine titled "Kicking Robots," spent months testing machines from Apptronik and Agility Robotics and trying to understand the gap between pitch and reality. He described the experience to Vox this week: he pushed a prototype robot as hard as he could, watched it stagger and throw its arms up to regain balance, and felt "an uncanny moment" when the machine trotted back and looked him in the face. Impressive, yes. But it was also a prototype worth hundreds of thousands of dollars, performing a task far simpler than anything a general-purpose domestic or industrial robot would need to handle.
"I think it's nearer to flying cars than it is to the chatbot side of things," Vincent said when asked to locate humanoid robots on the hype-versus-reality spectrum. He sees genuine capability advances but says the timelines being offered by companies like Tesla, which promises Optimus robots in homes soon, are simply not credible. The analogy to chatbots is doing a lot of work in this industry, and it is where Vincent's scepticism bites hardest. Large language models get things wrong regularly, but a wrong answer in a chatbot is an inconvenience. A robot that breaks one cup in ten while clearing the table is not acceptable. The error tolerance for physical systems is orders of magnitude tighter.
That said, not everything happening right now is pure hype. SAP's warehouse deployment is narrow but real: robots autonomously handling logistics tasks in a production environment, not a demo lab. The App Store model Unitree is pushing is architecturally interesting, even if it presupposes robots capable of reliably executing downloaded motion programs. It is the software layer being constructed while the hardware layer catches up.
The US-China contrast is becoming sharper. Vincent notes that Chinese companies are building faster and at larger scale, driven by demographic urgency: people over 60 are projected to be 30 percent of China's population by 2040, meaning the country needs robots for elder care and factory labor in a way that is structural rather than aspirational. US companies, by contrast, tend to pitch home robots as premium products for affluent consumers, a market with different tolerances and expectations. Tesla's Optimus is explicitly designed for domestic use. China's leading humanoid firms are building for warehouses and care facilities.
Colin Angle, who invented the Roomba and built iRobot into a household name, recently launched a new company with a revealing design decision: he explicitly rejected the humanoid form. His new product, "Familiar," is an AI-powered companion creature that communicates through meowing and body language, not speech or dexterous limbs. Angle's bet is that genuine connection, not humanoid dexterity, is the domestic market's real unmet need. It may be the most pragmatic reading of where the technology actually is: building what is achievable now, rather than what the pitch decks promise for next year.
The money flowing into humanoid robotics is real, the technical progress is real, and the enthusiasm is real. The App Store is here. The question is whether the robots it is designed for will be ready for it before the funding cycles run out of patience.