Chinese AI stocks have crashed. Does that tell us what will happen to AI stocks in the US?

Let's take a closer look at facts and data.

What happened in China (Chart 1): China's AI and chip index lost 40%, about $480B, in four months. In January, Zhipu and MiniMax became the first AI labs in the world to go public. Small investors asked for 1,800 times the MiniMax shares on offer. By May, Zhipu traded at 17 times its IPO price. Then it all reversed. On 22 Oct, MiniMax was worth less than at its IPO, 88% below its peak, Zhipu 77%.

Why?

1. China's economy: The government is cutting spending into an already relatively weak economy, and the property crisis goes on. That hits every Chinese stock.

2. Too many shares for sale: In January Zhipu and MiniMax were the only AI labs you could buy; scarcity set the price. Then the early investors were allowed to sell in July, more AI companies listed, and the price collapsed. That is just supply and demand, the same on every stock exchange.

3. The Chinese labs do not make much money (Chart 3): They give their models away for free. Users and talent came, revenue did not: China's six biggest AI labs together earn about one tenth of what OpenAI and Anthropic earn.

Now the US (Chart 2).

Reason 1 does not apply.

Reason 2 is already playing out. SpaceX raised $75B in June, the largest listing ever, and its shares have since fallen well below their peak. Anthropic reportedly wants to raise about $100B, more than every other US IPO of 2026 combined, except SpaceX. Its listing has been pushed back twice, from October to mid-October to "November". OpenAI filed in June and has no date; reports now say 2027. When sellers keep waiting, it's usually because they are not sure buyers will pay the price asked.

Reason 3 is where the US is still different but maybe not for long: Anthropic made $11.5B of revenue in a single quarter (with large losses). The Chinese labs have nothing like it. But their free models get better and cheaper every month, which could push down the prices OpenAI and Anthropic can charge as we all learn to leverage cheap models too. And the more usage shifts to cheap models the more the US and Chinese AI stocks markets may converge.

So, two lessons to remember from China:

- the first day of trading proves nothing, wait for the day insiders are allowed to sell

- the US labs have the revenue but what they have to prove is that they can keep charging for it while China gives it away. For now I think it comes down to their ability to innovate fast. Not great news for the AI doomers.

Analysis: NeuronVentures