Ashlee Vance on humanoid robotics: China dominates actuators, US software lead may not be enough

Summary

Ashlee Vance lands firmly in the Unitree camp when asked to choose between Figure at a $39 billion valuation and Unitree at $7 billion. The hardware supply chain is the core reason. Actuators, the motors that drive humanoid movement, are overwhelmingly manufactured in China, and every major non-Chinese robotics program depends on them. Tesla placed a roughly $700 million actuator order, which Vance reads as a signal that Elon Musk is planning significant near-term Optimus volume. Musk's position, relayed directly, is that Tesla prototypes actuators in China but intends to manufacture them domestically.

China's Hardware Lead vs. America's Software Advantage

China is making measurable progress on balance and movement mechanics, while the US retains a meaningful software lead. Vance believes China will eventually close that software gap, but the gap remains wide enough today that it is actively constraining the entire field. The thesis is that whoever solves dexterity and software together first wins, and right now no single player has both.

Boston Dynamics is not seen as a credible counter to Unitree. Despite producing technically impressive demos, the company has sold relatively little at scale beyond limited military contracts. Its ownership has changed hands multiple times, and its positioning in the humanoid race is unclear even to people close to the robotics industry.

Regulation Risk Is Real but Not Imminent

The DJI precedent is the relevant frame. Chinese drones embedded in US police forces and military units created a straightforward national security argument for restriction. Humanoid robots are not yet at that integration level, but the moment armed forces begin serious deployments, Unitree and its peers become obvious regulatory targets. There are roughly a dozen Chinese humanoid companies operating, which complicates any blanket restriction.

Consumer and Industrial Readiness

Current humanoid robots rate roughly 1 out of 10 as entertainment products despite the concept scoring a 10 out of 10. The primary technical failure mode is thermal: motors overheat under high-demand conditions like rapid punching sequences and freeze the unit. Repetitive, regulated tasks like laundry are more manageable precisely because the movement envelope stays narrow. Teleoperation has shown almost no meaningful advancement over the past decade in Vance's assessment, with demos from companies like 1X still suffering from the same instability issues that characterized the category years ago.

The longer-term bull case on humanoids is the DARPA Grand Challenge parallel. Autonomous vehicles looked broken at the first event, then made a step-change leap in the second iteration, and now Waymo units operate routinely in Austin and San Francisco. Vance frames humanoids as the technology most likely to follow that same arc over the next ten years, even if current progress looks negligible.

Other Hard Tech Callouts

New Limit, a longevity company backed by Brian Armstrong and run by Jacob Kimmel, is flagged as underappreciated, described as producing significant science output with a small team. Casey Handmer's company draws respect for the ambition of its mission, with the view that the scale of the potential outcome justifies the capital risk. Vance visited Helion's facility and called their new reactor among the most impressive pieces of room-sized hardware he has encountered, though the company declined to allow it to be filmed.

On quantum computing, Vance has covered the sector since a D-Wave story roughly 15 years ago, when general-purpose quantum computing was described as a few years away. His skepticism has only deepened. He notes that even PhD-level researchers in the field cannot consistently agree on whether a given system is actually functioning as a quantum computer, which he treats as a disqualifying signal for serious near-term investment theses.