Goldman Warnes of “Optimistic” Humanoid-Robot Capacity Build-Out as Supply Chain Awaits Actual Orders

Goldman Sachs’ latest supply-chain tour reveals an industry racing to scale production capacity for humanoid robots before any meaningful orders materialize—a development that underscores both bullish long-term sentiment and near-term uncertainty about when commercialization will truly arrive.Following meetings with nine Chinese suppliers between 3 and 6 November, Goldman analyst Jacqueline Du’s team found companies preparing capacity for anywhere from 100,000 to one million robot-equivalent units annually. That looks remarkably optimistic against Goldman’s own forecast of just 1.38 million global humanoid-robot shipments by 2035—suggesting suppliers are betting heavily on explosive growth that may or may not materialize on their ambitious timelines.The Capacity Race Without Orders

The disconnect is striking: most suppliers are “actively planning capacity both in China and overseas (primarily in Thailand, and less in Mexico),” yet “no company has yet confirmed sizable orders or definitive production timelines,” according to the Goldman report. The automotive supply chain, desperate for new growth engines amid EV-market maturation, appears eager to pivot into robotics—but the pivot is occurring largely on spec.Sanhua Intelligent Controls exemplifies the scale of preparation. The company has secured roughly 200,000 m² of land in Thailand—initially reserved for traditional auto parts—and purchased an additional plot of equal size specifically for humanoid-actuator assembly. Yet management sees “no need for large-scale production-line build-out immediately,” planning instead for “gradual ramp-up upon customer’s actual orders.”The company maintains it holds more than 50 % market share in actuator assemblies and will “continue to focus on one single leading global customer”—widely understood to be Tesla—rather than pursuing broad-based expansion at this stage.Aggressive Positioning, Fluid Timelines

Tuopu Group Technology is planning even more aggressively, with RMB 7–8 billion (US $1–1.1 billion) in total capex for humanoid-related capacity across Thailand, Mexico, and the United States. Its Thailand plant alone targets one million units annually, though management acknowledged “production planning remains fluid, with frequent revisions depending on client timelines and program evolution.”Translation: nobody really knows when volume production begins, but suppliers want to be ready when it does.The supply chain is also rapidly expanding product portfolios, evolving from single components to integrated modules. Minth Group is targeting RMB 5 billion in humanoid-related revenue by 2030, assuming 20–30 % market share, RMB 20,000 (US $2,800) content value per robot, and one million global unit shipments. The company has built a 10,000-unit-a-year capacity line for head-and-face assemblies, expecting commercial production in 1Q 2026.The Technical Arms Race

What is driving this pre-emptive capacity build is partly technical competitiveness. Companies are aggressively showcasing “rapid design-to-product turnaround” and “agile service as key comparative edge” to secure supply-chain position. Rongtai Health Technology claims it can deliver “from drawings to functional samples within 14 days,” while Shuanghuan Driveline has been sending samples to its North-American customer for “eight to nine rounds to evaluate parameters.”Dexterous hands remain a particular bottleneck. Zhaowei Machinery & Electronics prices its second-generation standardized dexterous hands at roughly US $17,000 each, expecting this to decline to US $7,000 per hand once mass production begins. The company anticipates RMB 100 million in dexterous-hand revenue for 2026, though current sampling orders remain “limited to several hundreds of modules.”The Reality Check

Goldman’s checkpoints for assessing whether a technology inflection point is imminent are revealing:

Tesla Optimus Gen-3 launch by February/March 2026.

Public disclosure of 2026 order/shipment targets by year-end 2025 or early 2026.

In other words, the supply chain is building capacity for volume production commonly expected in 2H 2026, but concrete evidence of demand remains elusive. Companies mentioned customers including Tesla Optimus, Agibot, Leju, XPeng, and others—but these are early-stage players who themselves are unproven in commercialization.The Goldman team maintains a “constructive” view on the long-run humanoid-robot trend while acknowledging the need to “monitor key robot-product performance and concrete end applications” before declaring an inflection point near. They hold Buy ratings on Sanhua H-shares, Shenzhen Inovance Technology, and Shuanghuan, with Neutral ratings on multiple others in the space.The broader message: China’s automotive supply chain is making a calculated bet on humanoid robots as the next growth frontier—but it is a bet being placed before the game has truly begun.