What I Learned Watching a Humanoid Robot Do Laundry

Welcome back to In the Loop, TIME’s new twice-weekly newsletter about AI. If you’re reading this in your browser, why not subscribe to have the next one delivered straight to your inbox? What to Know: My day with the robots This summer, I found myself in the strange position of watching a humanoid robot try to load laundry. It squatted beside a washer-dryer unit, reached with one hand into a laundry basket that it was holding with the other, and put some clothes into the drum. But twice in a row, it dropped a piece of clothing and couldn’t pick it back up. An engineer with a litter-picker grabbed the fallen cloth and sheepishly moved it behind the machine, out of my line of sight. But the robot’s limitations were too obvious to hide. Buzzy startup — I was visiting the headquarters of Figure AI, an audacious startup that hopes to one day be the biggest business on the planet. Last week, they announced the Figure 03, their newest humanoid, to great industry fanfare. The launch made it clearer than ever that the vast optimism of the AI boom is spilling over into robotics. Dozens of companies are now competing to make the world’s first viable humanoid robot, convinced that recent advances in AI have made what was once a pipe dream into achievable reality. Figure AI is now one of Silicon Valley’s buzziest—and most valuable—startups, having just raised $1 billion at a $39 billion valuation. Cover star — We put the Figure 03 on the cover of TIME last week, as part of our Best Inventions of 2025 issue. I wrote the story to accompany that cover. I see the purpose of the article as twofold. First, to inject some much-needed eyewitness evidence into the news cycle around Figure’s robots, which are not yet available for purchase. So far the public has only been able to assess their capabilities via nicely-shot YouTube videos and company blog posts. My reporting indicates that Figure’s robots are not quite as reliable as those media might have you believe. The robots are coming — But second, I wanted the piece to grapple with the possibility that humanoid robots might nevertheless be around the corner. Readers of this newsletter will probably be familiar with the AI scaling laws—the observation that an AI’s capabilities increase in line with the amount of (useful) data and computing power that it is trained on. For example, GPT-2 was pretty dreadful upon its release in 2019, but four years later, GPT-4 blew everybody’s minds. Point being: AI can gain new capabilities disarmingly quickly. Figure argues—convincingly—that its robots are on this trajectory. So, given the billions of dollars being poured into making robots better, we should seriously prepare for the possibility that humanoids might be entering the workforce in a matter of years, not decades. You can read my full Figure 03 feature here. Please feel free to email [email protected] and tell me what you think! Who to Know: Hock Tan, Broadcom CEO OpenAI continued its spree of deals with chipmakers on Monday, with the announcement that it would team up with Broadcom to design and deploy 10 gigawatts’ worth of AI accelerators. Broadcom has firepower in the world of AI. It works with big tech companies to custom-design proprietary chips; for example, it has a longstanding partnership with Google to develop the search giant’s Tensor Processing Unit—an AI chip that Google uses as an alternative to Nvidia’s GPUs. Now it looks like OpenAI is trying to get in on a similar deal. “By designing its own chips and systems, OpenAI can embed what it’s learned from developing frontier models and products directly into the hardware, unlocking new levels of capability and intelligence,” OpenAI wrote in a Monday blog post. For Hock Tan, Broadcom’s CEO, the OpenAI deal is just the latest milestone in a bull year that has seen his company’s stock rise more than 30% even before the jump caused by the OpenAI announcement. That makes Broadcom the seventh-most-valuable company in the world. AI in Action A stack of new AI bills was signed into law by California Governor Gavin Newsom on Monday, including SB 243, a bill that will bring in measures designed to protect children from “companion” chatbots. The law, drafted by state senators Steve Padilla and Josh Becker, will come into effect on Jan. 1, 2026, and requires companies offering companion chatbots to implement age verification and disclose to the state their protocols to address suicide and self-harm. “These companies have the ability to lead the world in innovation, but it is our responsibility to ensure it doesn’t come at the expense of our children’s health,” Padilla said in a speech just before the bill’s passage. What We’re Reading America’s future could hinge on whether AI slightly disappoints — by Noah Smith on Substack There has been growing debate about whether we’re in an AI bubble. Yet most AI researchers report that AI capabilities are on an upward trajectory, andcompanies like OpenAI, Microsoft, Google, Amazon, Meta, and Nvidia plan to continue investing hundreds of billions in data center buildouts for the remainder of the decade. That optimism might put investors’ short-term fears to rest, even amid lingering concerns over whether AI will bring enough returns on investment to justify the increasingly debt-fueled splurge. But economist Noah Smith wades in with a sobering observation: “When we look at the history of industrial bubbles, and of new technologies in general, it becomes clear that in order to cause a crash, AI doesn’t have to fail. It just has to mildly disappoint the most ardent optimists.”