China tries to kickstart a bubble in humanoid robots

As well as the AI bubble in the US, there’s a separate AI bubble trying to get going in China. It started as the Chinese real estate bubble trying to pivot to a new field. So it first showed up as a massive data centre buildout. You’re seeing a wave of Chinese chatbot models because the companies…

As well as the AI bubble in the US, there’s a separate AI bubble trying to get going in China. It started as the Chinese real estate bubble trying to pivot to a new field. So it first showed up as a massive data centre buildout. You’re seeing a wave of Chinese chatbot models because the companies saw how well OpenAI got venture capital, and they wanted in. So they’re putting out their models in the hope of IPOing on the Shanghai or Hong Kong stock markets before the worldwide AI push runs out of steam. Why did Moonshot put out the Kimi 3 model as open-weights, and offer API access cheap? Because Moonshot’s planning to IPO in Hong Kong in 2027, and Kimi 3 promotes that IPO. [Bloomberg, archive; KrASIA] But the Chinese AI bubble is not just the chatbots. There’s also the humanoid robots. There’s two big problems with humanoid robots: they’ve got no use case and they don’t work. But the Chinese Ministry of Industry and Information Technology has a bee in its bonnet. The Ministry demands: [SCMP, archive] By the end of 2026, key humanoid robot products will complete application verification and regular deployment in a number of representative scenarios, entering ‘work mode’. China ran a robot Olympics to showcase the machines. Mostly it was a source of robot prat fall videos. [Bluesky, video] The dancing and running robot demos are impressive, the same way ChatGPT was impressive. But “impressive” does not mean useful for real work. So the hype is made of ”could” and ”might” and ”soon”, which are all words that mean “doesn’t actually”. The robots have done well for money at least. Unitree IPOed in Shanghai with a 600% pop on its first day on the markets. At the end of that day, Unitree was valued at over $50 billion. The offering was 5,500 times oversubscribed from retail investors. [FT, archive] Unitree’s stratospheric valuation is on claimed sales of just 5,215 robot bodies in the past year across all of China. Most of those were sold to universities as research platforms. It might all be a bit bubbly. [FT, archive] The market for the robots is approximately 100% financial kayfabe. It’s circular financing with state enterprises.: [FT, archive] China’s humanoid robot makers are generating much of their revenue from selling machines to government-backed training centres — which then collect and sell training data back to the robot makers. … The centres, often co-funded by local governments and robot makers, buy the machines, generate training data and sell the data to the robot makers to improve their technology. This government subsidy means a flood of humanoid robot startups: Nearly 370 have been established in the past two years, and more than 50 of them have listed or are preparing to go public. China has 97% of the market for “humanoids”! That’s because it’s a fake market. The actual market barely exists — it’s novelty robots at sales launches and similar. If they don’t break embarrassingly. [Bloomberg, archive] Last year, analysts at Morgan Stanley said humanoid robots would be a $5 trillion market! This comes after Morgan Stanley’s previous predictions of the Metaverse as an $8 trillion market and luxury NFTs as a $56 billion market. So I guess the robots are screwed then. [Morgan Stanley, 2025; Forbes, 2021; Business Insider, 2021] Video — Podcast

Source: Pivot to AI — Published August 31, 2026 at 22:44 UTC — Category: Business