VERTEBRATE.ai

At its 2026 Innovation Day, the Shanghai-based humanoid robot maker AgiBot introduced Worken, a system designed to do something the industry has mostly talked around until now: measure, in hard numbers, how much actual work a robot performs on a job. The announcement came alongside a separate but related data point reported by Global Times — AgiBot and a cluster of Chinese humanoid robot peers are posting revenue growth that is outrunning even the breathless pace of their venture funding rounds, an early sign that these machines are leaving the lab and entering the general ledger.

Worken itself was light on technical detail in AgiBot's unveiling, but its purpose was plain: give customers, investors, and AgiBot's own engineers a standard yardstick for robot productivity, rather than relying on demo reels and press-day choreography. For an industry that has spent the better part of a decade selling potential — footage of a humanoid folding a shirt, pouring a drink, stacking a box — a formal metric for output is a notable pivot. It suggests AgiBot wants its machines judged the way factory equipment or delivery fleets are judged: by throughput, not by applause.

That pivot matters because the humanoid robot sector has long had a credibility problem baked into its business model. Billions of dollars in venture capital have flowed toward companies promising general-purpose robots capable of working alongside humans, but proof of sustained, paid deployment has been thin. Revenue growth outpacing the VC cadence — as Global Times reports is happening at AgiBot and its rivals — is the first real signal that paying customers, not just funding rounds, are driving the numbers. Pairing that growth with a tool like Worken looks like an attempt to make the case stick: here is the revenue, and here is the system that shows what that revenue is actually buying in terms of labor performed.

The stakes extend beyond one company's marketing strategy. China's humanoid robot makers are racing to establish themselves as the default suppliers for warehouses, factories, and eventually homes, in direct competition with American and Japanese rivals who have made similar promises with less commercial traction to show for it. A credible productivity metric, if adopted industry-wide, could reshape how buyers compare robots the way horsepower or miles-per-gallon once reshaped how people bought cars — turning a marketplace built on hype reels into one built on comparable performance data. For workers and employers watching this technology edge toward ordinary workplaces, that shift from spectacle to spreadsheet may be the more consequential story: it is usually the moment when a technology stops being a demo and starts being a line item.