A humanoid robot ran 100 meters in 9.39 seconds at the opening ceremony of the second World Humanoid Robot Games in Beijing last week, beating Usain Bolt's 16-year-old world record of 9.58 seconds. Days earlier, in an unofficial trial, a robot called Lightning, built by Chinese phone maker Honor, went even faster: 9.32 seconds, hitting a peak speed of 14.5 meters per second. In the standing high jump, a robot from Beijing-based X-Humanoid cleared 2.88 meters, past the 2.45-meter human mark Cuban athlete Javier Sotomayor set in 1993.
"The major trend will be a shift from pricing, evaluating, and managing risk around human operators to system-centric operators or system-centric liabilities," says Chris Raimondo, insurance consulting leader for the Americas at EY.
That's the shift the sprinting robots obscure. More than 2,000 robots from 666 teams competed across the five-day event, held at the National Speed Skating Oval built for the 2022 Winter Olympics, up from roughly 500 robots and 280 teams at last year's inaugural games.
Sprinting is the easy part, and it isn't where the insurance industry's exposure sits. Of the 51 total events, roughly 40% required a robot to complete a work-style task, connecting cables, sorting warehouse packages, charging an EV, taking a restaurant order, with no human operator standing by to intervene.
A machine that can locate a cable at an awkward angle, grip it correctly and plug it in without knocking over the shelving next to it is solving a harder problem than running in a straight line. It's also a closer analogue to the warehouse, retail and facilities exposures already sitting on carriers' books.
This isn't a hypothetical for the US market. Raimondo told Insurance Business earlier this year that so-called "physical AI," autonomous vehicles and humanoid robotics together, could prove more disruptive to carriers than generative AI has been, because it forces a shift from pricing human operators to pricing systems. Liability on a single incident, he said, could eventually span the hardware manufacturer, the AI software platform and the commercial owner at once.
Brace yourself for a future where robots go rogue. Unitree H1 humanoid suddenly moved wildly due to a coding error. pic.twitter.com/ZYYFY93RPk
— Interesting World (@_fluxfeeds) May 2, 2025
The same reallocation touches workers' comp. In an April interview with Insurance Business, Raimondo said claims built on eyewitness accounts and physical evidence will increasingly give way to sensor logs, OEM data and software diagnostics, a shift that could push workers' comp products to resemble broader commercial package policies rather than a standalone line. Traditional injury claims may decline as robots absorb repetitive, physically demanding tasks. System failure, cyber vulnerability and business interruption are expected to take up more of the space claims teams once spent on eyewitness accounts.
Executives at the games weren't shy about the gap between spectacle and utility. Yu Chao, chief executive of competing firm Lumos Robotics, put it plainly: "Simply running and jumping does not improve efficiency." Hua Rong, chief marketing officer at rival firm Zeroth, framed the real test as commercial: whether the product can solve users' problems after it's sold, not how it performs in a choreographed heat.
The labor-market data backs up that skepticism about humanoids specifically, even as it confirms automation overall is accelerating. The World Economic Forum's Future of Jobs Report 2025, based on a survey of employers representing more than 14 million workers across 55 economies, projects robots and automation will displace roughly 5 million more jobs than they create globally by 2030. That's a smaller effect than the net 2 million jobs the WEF expects from AI and information-processing technology. Close to 39% of workers' core skills are expected to change by 2030 across all trends in the report.
Most of that displacement is still coming from conventional industrial robotics, not humanoids. The International Federation of Robotics counted 542,000 new industrial robot installations worldwide in 2024, more than double the figure from a decade earlier, pushing the global operational stock to roughly 4.66 million units. China accounted for 54% of that year's new installations alone.
Humanoids remain a rounding error by comparison, and forecasters don't agree by how much. Goldman Sachs Research put the global humanoid market at $38 billion by 2035, a sixfold jump from its own prior-year estimate. Other analysts size today's market at $3 billion to $6 billion, with 2035 projections running as high as $50 billion. Treat any single figure here with some skepticism. This is a young market, and credible estimates currently disagree by an order of magnitude.
The clearest example of displacement so far isn't a humanoid at all. Internal Amazon documents reviewed by The New York Times, suggested the company could avoid hiring more than 600,000 US workers it would otherwise need by 2033 as conventional warehouse automation scales.
Amazon pushed back on the framing, telling reporters the leaked documents reflected "one team's perspective" rather than company-wide strategy. It's a payroll and workers'-comp story playing out in ordinary fulfillment centers, years before humanoids do the same work reliably and unsupervised.
There's a new wrinkle for commercial clients that import robotics hardware. In late July, the Federal Communications Commission added foreign-produced advanced robotic devices, including humanoids and quadrupeds, to its "Covered List," blocking new models from receiving the FCC equipment authorization needed to import, market or sell them in the US, absent a conditional approval from the Department of War. The listing is formally "country neutral," turning on where a device is produced rather than the manufacturer's nationality, but it lands hardest on Chinese suppliers, chiefly Unitree, AgiBot and UBTech, who together ship somewhere around 80% of the world's humanoid robots. Equipment already authorized or in use isn't affected, and manufacturers can apply for conditional approval before January 1, 2028.
For brokers advising manufacturing, logistics or construction clients, that's a supply-chain and compliance question worth adding to the usual safety and liability review. Where was the device built? Is its model still eligible for FCC authorization? Does the client's technology E&O or contractual risk-transfer language account for a supplier being cut off mid-contract?
Worth checking, too, is what's already been quietly written out of a client's existing coverage. A nationwide review of state insurance filings found carriers had filed generative-AI exclusions across standard commercial lines contractors carry in every state but one, built off six standard forms ISO published in July 2025. Robotics hardware sits next to, and sometimes inside, the same AI-driven operations those exclusions target. A policy renewed even a year ago is worth a second look before a client puts a humanoid or an AMR fleet on the floor.
Robots aren't coming for anyone's job on the timeline the viral clips suggest. But the risk pool underwriters have priced around human operators for a century is starting to fragment across manufacturers, software vendors and commercial owners, in warehouses and factories today, and potentially in humanoid form within the next underwriting cycle or two. Sprinting robots make for good video. A machine that can plug in a cable, unattended, on its five-hundredth try is the one worth pricing for.
