A humanoid-shaped bubble might be forming in China’s tech scene — and this time the warning came not from pundits or short sellers but from the National Development and Reform Commission (NDRC), the country’s top economic planner. At a Beijing press briefing on November 27, the commission’s spokesperson, Li Chao, told reporters that the sector needs to “balance the speed of growth against the risk of bubbles,” flagging a rush of investment into humanoid robots even though practical use cases remain thin.

Li’s blunt assessment leaned on a stark number: more than 150 companies in China are now building humanoid robots, and “more than half” of those are startups or businesses jumping in from adjacent industries. That kind of headcount — a scramble of firms chasing the same futuristic promise — is exactly what policymakers worry can turn a hot sector into an overheated one, with copycat products, compressed R&D budgets, and a fragile market when the next funding cycle tightens.

The timing of the warning matters. Beijing has been actively positioning “embodied intelligence” — the class of AI that gives physical machines perception and motor control — as a strategic area for the nation’s future economy, handing it high-level backing and a place on policy roadmaps earlier this year. That official push helped spark a wave of interest from manufacturers, suppliers, and VCs hoping to catch the next big industrial AI play, but it also helped create the conditions for hype to outpace utility. In short, policy signals encouraged scale, and scale now risks encouraging sameness.

What keeps NDRC officials up at night is not the robots themselves but the pattern that often follows rapid growth: lots of lookalike machines rolled out to market because founders want press and investors want traction, not because there’s a validated business case. Li warned against “highly similar” models flooding shelves — a scenario that drains capital from genuine research, squeezes margins, and leaves the industry vulnerable to a sudden re-pricing when investors pivot away. It’s a story investors and entrepreneurs know well from past tech cycles, and authorities are clearly trying to steer the outcome toward consolidation and substantive progress rather than spectacle.

You can see the balancing act at play in concrete examples. Chinese firms have unveiled humanoid prototypes that grab headlines — Xpeng’s “Iron” series among them — but the gulf between polished demos and daily, revenue-generating deployments remains wide. That gap is what separates corporate R&D milestones from an actual market that will pay for serviceable robots in logistics, manufacturing, elder care, or hospitality. Until those repeatable use cases exist at scale, money poured into publicity-first products risks evaporating.

The broader context matters too: the humanoid moment is arriving amid widespread jitters about an AI funding reset. When core AI models and startup valuations are being scrutinized, a capital-intensive hardware play that depends on long development cycles and expensive supply chains becomes especially exposed. If venture and corporate budgets retrench, the humanoid space could see smaller players fail quickly and larger players consolidate or refocus on software and components rather than full humanoid builds.

So what would a healthier path look like? For regulators and industry watchers, the answer is familiar: push firms toward differentiation, make R&D funding conditional on demonstrable progress, and encourage partnerships that move prototypes into repeatable commercial pilots. Investors will also need to tighten their playbooks — more diligence on unit economics and fewer bets on headline-grabbing demos. And for the startups themselves, the near-term survival play is simple but hard: stop competing to be the flashiest robot and start proving where a humanoid actually saves time or money.

China’s warning is both a caution and a nudge. Officials want the embodied intelligence industry to succeed — they’ve signaled that success matters to the country’s technological ambitions — but they also want that success to be sustainable. Whether that nudge will slow the stampede or merely reroute it remains to be seen; for now, the message is clear: build faster if you must, but don’t forget to build differently.