China wants 11 million humanoid robots in its factories. Last year it sold 12,000.
Shanghai based Agibot livestreamed six factory days to back China's Five Year Plan, which lists humanoid robots as a top 10 strategic industry.
Shanghai based Agibot livestreamed six factory days to back China's Five Year Plan, which lists humanoid robots as a top 10 strategic industry.
China's latest Five-Year Plan has elevated humanoid robots to one of ten strategic industries the central government wants to dominate within the next decade (The Independent). A six-day livestream last month from Shanghai-based Agibot put a face on that ambition: more than 64 hours of quality-control inspections and materials handling at a claimed success rate of 99.99 per cent (The Independent). The picture the video paints, however, is not the picture the industry's own analysts are willing to defend.
The disagreement is roughly three orders of magnitude wide. Barclays has projected that around 11 million humanoid robots will be working inside Chinese factories by 2035 (The Independent). Morgan Stanley counted closer to 12,000 humanoid units sold across all of China's makers in 2025 (The Independent). Neither forecast is fake; both come from teams with reputations to protect. What they actually disagree about is how fast the gap closes.
The livestream does narrow one part of that question. It shows that the work Agibot's robots were designed for, repetitive inspections and moving parts between stations, can run for days at a time with only the occasional intervention. Agibot senior vice president Dr Yao Maoqing framed the event as a move from "demos" to "deployment" (Agibot). The company also announced that its 15,000th embodied robot had rolled off the production line (Gasgoo), a cumulative manufacturing milestone that is separate from the deployed count.
Two things keep the demonstration modest rather than definitive. First, the success rate is a company figure from a company-run livestream, not a third-party audit. The Independent's headline rounds the same number to "about 99 per cent," which is the kind of imprecision that happens when one number stands in for many. Second, the scope was deliberately limited. The robots inspected and moved components. They did not assemble them. Agibot and Longcheer Technology have separately announced what they call the world's first deployment of "embodied AI" in a consumer-electronics precision manufacturing line (Agibot), a different and somewhat harder claim.
State media coverage is part of the story but not independent validation of it. Xinhua carried the demonstration (Xinhua) and the State Council Information Office echoed the milestone in its own English coverage (SCIO). That amplification tells you the campaign fits the industrial-policy line Beijing wants to project. It does not tell you the robots are working today in the numbers the policy assumes they will by 2035.
What changes the picture in the next 18 months is whether independent deployment data starts tracking Barclays' curve or Morgan Stanley's. If monthly installation reports from tier-one Chinese manufacturers begin climbing into the tens of thousands, the livestream model, with its controlled conditions, continuous monitoring, and deliberately narrow scope, will look like the template that proved the skeptics wrong. If the next two quarters still read in the low thousands, the Five-Year Plan's mass-deployment target becomes harder to defend without redefining what "deployment" means. Either way, the next sales round, more than the next livestream, is the number worth watching.