J.P. Morgan expects United States humanoid robot output to reach between 300,000 and 400,000 units by 2030, and about 1.5 million units by 2035, as the industry shifts from demonstrations to commercial deployment. The 400,000 figure that appears in many headlines is the top of that range rather than a single forecast. The estimate, reported by Reuters on November 12, 2026, comes as the industry moves beyond demonstration robots toward factory and warehouse work.
The Forecast in Context
The bank's view is that demand is coming from manufacturing, logistics and warehousing, where labor shortages and rising production capacity are pushing companies toward automation. The note describes the industry as moving from demonstrations to commercial deployment, which makes production capacity the key question.
The gap between that outlook and current plans is wide. According to the report, most U.S. humanoid makers other than Tesla are targeting only 10,000 to 30,000 units a year. Reaching the bank's 2030 range would therefore require a steep ramp, a small number of very large producers, or both. That tension is central to how the forecast should be read.
Reuters' coverage of the note names Tesla Optimus, Agility Robotics' Digit, Boston Dynamics' Atlas, Figure AI and Apptronik's Apollo as the platforms in play. Tesla is the exception to the 10,000-to-30,000 target range, which is why the bank's upper bound depends so heavily on that company's plans.
Bottlenecks Are the Real Constraint
The report's central argument is that the limiting factor is hardware and compute rather than customer interest. J.P. Morgan says progress is supply-constrained, citing "hands, actuators, the brain, and production capacity" as the binding limits. Deployment is also held back by the availability of AI training data and computing power.
In practice, that means a shortfall in any one of those areas can hold back the others. The "brain" refers to the AI models that let a robot perform varied tasks, and the bank ties it to the availability of training data and compute.
Economics and the Labor Case
The most striking claim in the note concerns cost. J.P. Morgan estimates that humanoid robot costs are nearing parity with labor costs, and that multi-shift use could save about $100,000 a year per customer. The savings figure assumes robots run across multiple shifts, so reliability over long working hours would matter as much as the sticker price.
The report also points to a shift in where value will accumulate. It expects the market for actuators, motors and vision systems to approach $30 billion by 2035, and it favors suppliers including Timken, Robotis, Leader Drive, Orbbec and Cognex. Over time, the bank expects more of the value to move toward robot makers and the integrators who fit robots into real operations.
Trade Policy Is the Wild Card
The bank also flagged trade policy, saying the administration's national security finding on imported robotics could lead to tariffs or other measures that boost domestic production. That finding, signed by President Trump on October 28, did not itself impose tariffs, and AJOT reports that Trump said he could still impose them later. AJOT adds that officials were ordered to negotiate import levels with foreign governments and that Commerce and the U.S. Trade Representative must report back within 180 days. Separately, the Federal Communications Commission barred imports of new models of Chinese robots in July, while allowing existing humanoid and quadruped models to continue entering the country.
For U.S. makers, that combination could favor domestic production. For foreign suppliers, it adds uncertainty about where products can be built and sold.
The policy debate is not settled. The Association for Advancing Automation urged the president in October 2025 not to impose robotics tariffs and to develop a national robotics strategy instead, according to AJOT's account. The Commerce Department's own position, as summarized by AJOT, is that the U.S. "relies on foreign imports for most types of robotics and industrial machinery." That dependence is the reason the government is weighing tools that could reshape supply chains, and it is also why the humanoid forecast and the trade talks are tied together.
Why It Matters
Humanoid forecasts have varied widely, so this one stands out because it comes from a major investment bank that also has a view on suppliers. If the 2030 range is anywhere near right, the industry would need to move from pilot fleets to a scale that current U.S. makers, most targeting 10,000 to 30,000 units a year, have not yet reached. Component makers would see demand that today's supply chains are not built to meet.
The next checkpoints are production announcements from U.S. makers, the trade talks that could decide where factories are built, and whether labor-parity claims hold up in the field. Readers should watch the gap between announced capacity and actual shipments, since that gap will tell whether the forecast is a plan or a prediction.
