Humanoid robots are being judged on their demo videos, but the real contest is over actuators and motors — the parts that make up roughly 40–60% of the build cost and that, according to Robison's reporting corroborated in part by McKinsey, the US sources nearly or almost…
The humanoid robot industry has a $40,000 problem, and it lives in a motor.
Walk past the demo videos of robots folding laundry or striding through warehouses, and the question that decides whether any of these machines ship at scale is unglamorous: who makes the actuators. On the latest episode of Core Memory, hosts Kylie Robison and Ashlee Vance argue that this is the contest that actually matters — decided not in research labs or on launch pads, but in the supply chain for the small motors and gearboxes that move every joint in every humanoid robot (The Space Race Is So Back, EP 76, Core Memory).
Their framing is concrete: according to Robison's reporting, corroborated in part by McKinsey analysis, actuators and motors make up roughly 40–60% of a humanoid robot's build cost, and the US sources them nearly or almost entirely from China. That concentration is the load-bearing fact. Industry analysts at McKinsey have separately estimated the figure at 40–60% of humanoid build cost. The humanoid race, as they describe it, is not really a software race or even a design race. It is a manufacturing race — and right now the United States is buying most of the critical components from the country it is trying to out-build.
The supply concentration is not a secret inside the industry, but it has been easy to miss from outside, partly because the public conversation has been captured by the polished demos. Robison's underlying reporting identifies two US startups trying to onshore humanoid actuator production: Westmag and Atlas Motion Systems.
Westmag, co-founded by David Hansen and Jordan Sanders in South San Francisco, has raised an $11 million seed round led by Andreessen Horowitz, with Founders Fund, Lux, and Menlo Ventures participating. The company is building electric motors and actuators in the US using what Hansen calls a "dumb-guy, smart-guy" approach: don't try to improve the motor design, just commit to mass-producing it. The thesis is that at sufficient volume, the cost of almost anything drifts toward its raw material cost, and the US labor premium shrinks toward nothing. Hansen and Sanders have already partnered with high-volume customers — they won't name them publicly — and are targeting tens of thousands of motors per month by year's end. "By the end of this year we'll be in the tens of thousands of motors per month," Sanders says, only for Hansen to interject: "Which is not enough."
Atlas Motion Systems is the second startup in Robison's reporting. Co-founded by Tom Baron, Christian Mochen, and Carlo Dela Rosa — veterans of MITRE and the defense startup Mach Industries — the company is pursuing a different approach to the same problem. Baron, who spent years as a defense prototyper working across government agencies, argues that the real barrier isn't design but the entire supplier ecosystem that the US lacks. The Atlas team is building what Baron describes as the foundational capability the US needs to stop buying Chinese actuators by default.
The policy response is already forming. The Guarding the U.S. Against Adversarial Robotics Dominance (GUARD) Act was introduced by Rep. John Moolenaar, who chairs the House Select Committee on China, alongside Reps. Jay Obernolte of California and Jennifer McClellan of Virginia. The bill targets Chinese-made humanoid and quadruped robots — including Unitree, the Chinese maker of robot dogs that has been selling into the US market — on national security grounds, citing backdoor risks that could allow the People's Liberation Army remote access. The bill's sponsor framing compares the threat to the plot of the 1984 film Red Dawn. (Primary Congressional source for bill number and current status should be pulled from Congress.gov before final publication.)
What matters for the actuator story is that the same political energy that produces a Unitree restriction is the political energy that decides whether a US actuator startup gets defense procurement, Department of Commerce industrial-policy support, or a Customs ruling that reclassifies Chinese motor imports. The policy lever is real. What is not yet clear is whether it will move fast enough to matter inside the 18-to-36-month window the hosts describe for the next generation of humanoids.
The space race material in the same episode works as a useful parallel, not as the lead. Vance and Robison discuss a recent Blue Origin pad and vehicle explosion as evidence that the US lunar program is fragile, contrasted with the relentless launch cadence coming out of China. (The specifics of the explosion — vehicle identity, date, payload, FAA status — are not available in the public Core Memory transcript excerpt and would require primary sourcing before precise attribution.)
The honest read of the hosts' argument is that this is a ticking clock, not a victory lap. The US has the design talent and the venture capital. What it does not yet have is the production line. Until that changes, every humanoid that rolls out of an American lab is, in a literal sense, a tribute to a Chinese motor.
What to watch in the next reporting cycle: