JPMorgan published a forecast on November 12 projecting that US humanoid robot production could reach 300,000 to 400,000 units by 2030, climbing to roughly 1.5 million units by 2035. The bank also estimates the market for the components that go inside these machines — actuators, motors and vision systems — will approach $30 billion by 2035. The report names Tesla's Optimus, Agility Robotics' Digit, Boston Dynamics' Atlas, Figure AI's robots and Apptronik's Apollo as the makers most likely to drive that volume, alongside component suppliers including Timken, Robotis, Orbbec and Cognex.
The headline number is eye-catching, but the more useful part of the report is what JPMorgan says is actually constraining growth. It is not customer appetite. It is a narrow set of hardware bottlenecks: dexterous hands, actuators, the "brain" (compute and training data), and raw production capacity. Outside of Tesla, most US humanoid makers are currently targeting just 10,000 to 30,000 units a year — a fraction of the 2030 forecast, which means the gap between today's output and the bank's projection has to be closed by supply chain maturity, not by finding more buyers.
Why the Supply Side, Not Demand, Is the Story
For two years the humanoid robot narrative has centered on splashy demos and pilot deployments — a robot folding laundry here, one on a car line there. JPMorgan's report reframes the conversation around the unglamorous plumbing: can anyone actually build these machines at volume. Dexterous hands remain a particular choke point because they require dozens of small, precise actuators packed into a human-sized housing, a manufacturing problem closer to watchmaking than to welding robots. That is also why the bank flags actuator and motor suppliers like Timken and Robotis as being as important to the humanoid story as the robot makers themselves.
JPMorgan also points to a roughly $100,000 annual savings per customer when a humanoid can run multiple shifts, which is the economic case warehouse and factory operators are actually underwriting pilots against. That figure matters because it shifts the industry's proof point away from technical demos and toward a straightforward payback calculation that procurement teams can run.
The Competitive Field and the Policy Wildcard
The five companies JPMorgan names span very different bets: Tesla is pursuing in-house vertical manufacturing at a scale no rival currently matches, while Figure, Apptronik and Agility depend more heavily on outside component suppliers and warehouse or logistics customers such as GXO. Boston Dynamics, the most established name, is still treating Atlas as a research and control-software platform rather than a mass-production line. That split suggests 2030 output will likely be concentrated in a handful of companies rather than spread evenly across the sector.
The report also flags a less technical risk: the Trump administration is reviewing imported robotics and industrial machinery on national security grounds, a review that could push costs higher for companies relying on foreign actuators and vision components, or alternatively accelerate US domestic production incentives. Either way, JPMorgan's 2030 number is less a prediction of certain demand and more a ceiling defined by how fast hands, brains and factories can be built — the kind of constraint that has derailed optimistic robotics forecasts before.
Source: Investing.com (Reuters)