Shares priced at just over 150 yuan opened at 1,100 yuan. The Hangzhou robotics company, known for robot dogs and back-flipping humanoids, briefly commanded a valuation of roughly $66 billion. Retail subscription orders exceeded 7 trillion yuan.
It was an extraordinary vote of confidence in the future. It was also an extraordinary price to pay for it.
Ten weeks later, Unitree shares have fallen roughly halfway from their debut level. Across China’s AI and semiconductor sector, the damage is broader: the STAR50 has lost about 40% since July, erasing approximately $480 billion in market value.
For American investors, the question is whether the same financial assumptions underpinning China’s boom have become embedded in their own portfolios.
My view: China offers a warning about the price of AI optimism. Treating it as a precise forecast for Wall Street would go too far. Ignoring the similarities would be equally careless.
The First Casualty Is The AI Premium
China’s selloff has several identifiable causes. Stretched valuations, tax-related selling and the prospect of tighter American restrictions on Chinese data-center equipment have all contributed.
But those triggers landed on a market already vulnerable to disappointment.
Even after its decline, the STAR50 trades at roughly 100 times trailing earnings—more than three times the Nasdaq 100’s multiple. At that price, investors need years of exceptional execution to justify what they have already paid.
Concentration compounds the risk. According to the Goldman Sachs estimates cited in the market discussion, AI-related stocks account for 42% of China’s market value and 57% of trading.
When one investment theme occupies that much space, selling spreads easily. Investors reducing exposure to a crowded trade can pull down strong businesses alongside speculative ones.
The market begins by rewarding almost every company associated with AI. Eventually, it starts asking which companies can earn enough to support the valuation.
That transition is expensive for investors who bought the label.
Beijing Could Slow The Boom. It Could Not Supply The Earnings.
China’s response also challenges the idea that policymakers can reliably engineer a gentle end to speculative excess.
During the 2015 crash, Beijing mobilized state funds, restricted selling by large shareholders and suspended new listings. This time, state-backed investors sold into rallies, attempting to restrain enthusiasm before it became more dangerous.
The subsequent decline exposes the limits of that approach.
Authorities can influence liquidity and trading incentives. They have much less control over the revenue a business generates or the return it earns on new investment.
This distinction matters beyond China. Government support can accelerate an industry’s development. It does not guarantee attractive returns for shareholders buying at elevated prices.
An industry can be strategically essential and still be a disappointing investment.
Wall Street Faces A Cash-Flow Test
American technology companies enter this debate with substantial advantages. Microsoft, Alphabet and Amazon operate established businesses, serve enormous customer bases and generate real revenue from cloud services.
Those strengths deserve weight. So does the scale of their spending.
Alphabet expects capital expenditure of $195 billion to $205 billion this year. Amazon’s spending is expected to reach approximately $220 billion.
The investment case depends on demand growing fast enough—and remaining profitable enough—to justify that infrastructure.
Investors therefore need more than evidence that customers are using AI. They need evidence that usage produces durable economic returns.
A company can report rapid cloud growth while committing so much additional capital that the benefit to shareholders remains uncertain. Revenue growth and investment returns are related, but they are separate measures.
That is why the next earnings reports matter. Investors will be looking for clearer answers on capacity utilization, margins, customer demand and the path from spending to cash generation.
AI Can Transform The Economy And Disappoint Shareholders
The strongest argument against fears of an AI bubble is that the technology is useful.
It is also an incomplete argument.
A transformative technology can create enormous economic value while delivering poor returns to investors who paid too much. Competition can pass much of the benefit to customers. Infrastructure can become cheaper. Early leaders can lose their advantage.
The railway boom offers a familiar historical parallel: investors financed networks that outlasted many of the fortunes committed to building them.
AI infrastructure could follow a similar pattern. Data centers may remain valuable long after the valuations attached to their owners have reset.
My assessment is that the central investment risk lies in how much future success is already included in today’s share prices.
At a modest valuation, a business has room for delays and mistakes. At an exceptional valuation, even a successful business can disappoint.
China Is A Stress Test, Not A Timetable
There are good reasons to resist a direct comparison between Shanghai and New York.
The markets differ in investor composition, financing conditions, government intervention and exposure to geopolitical restrictions. American cloud platforms also have commercial foundations that many speculative AI businesses lack.
A 40% Chinese decline does not establish that a comparable American decline is coming.
It does, however, reveal how quickly enthusiasm can weaken when investors begin questioning the timing of returns.
That timing deserves particular attention. A company may eventually earn an attractive return on AI infrastructure and still face years of pressure before it gets there. Share prices can adjust much faster than business models.
The practical question for investors is therefore specific: what happens to this valuation if monetization takes two years longer than expected?
If the investment case cannot withstand that possibility, it depends heavily on an unusually favorable schedule.
The Earnings Miss Is Only One Warning Sign
A major earnings miss could trigger a broader repricing in American technology stocks. But investors should also watch for quieter signs of strain: rising capital expenditure, weaker cash generation, pricing pressure or customers taking longer to expand their commitments.
Any one of those developments could make the market reconsider how much it is willing to pay for future growth.
China’s crash does not settle the debate over AI’s potential. It shows that confidence in the technology offers limited protection against an excessive purchase price.
For Wall Street, the lesson is straightforward: keep evaluating the business after becoming convinced by the technology.
The future can arrive—and still fail to justify what investors paid for it.