The S&P 500 is still in a bullish trend. The Nasdaq is still near record highs. And one of Wall Street’s biggest banks is quietly telling clients to be careful.
JPMorgan’s technical strategist said this week that the AI trade is starting to look like something he has seen before. The last time it looked like this, the market crashed.
JPMorgan’s warning about AI stocks
Jason Hunter, JPMorgan’s chief technical strategist, told clients the current divergence in AI stocks mirrors what happened in the months before the dot-com bubble burst in 2000, according to CNBC.
His note, published Aug. 21, came as the S&P 500 had recently hit a fresh all-time high of 7,816 but was sitting well below its next resistance range of 7,909 to 7,935.
Hunter said the index remained above its support zone of 7,521 to 7,620 but flagged that conditions could deteriorate quickly heading into September.
The split he is pointing to is specific. AI hardware stocks have been on a tear. The Philadelphia Semiconductor Index is up 87% this year and logged its best-ever quarter in Q2. Memory stocks have run even harder. The Roundhill Memory ETF is up 141% since its April inception, according to 24/7 Wall St.
Related: Jane Street turns bullish on volatile AI stock
The companies spending the most on AI are a different story.
Meta is down 5% this year. Microsoft is down 18%. The hyperscalers, the companies writing the biggest checks for AI infrastructure, have been punished by investors asking whether the returns will ever justify the spending.
Hunter has seen that kind of split before. In 1999, communications equipment suppliers surged while the companies making heavy capital investments crashed from peak valuations. The dot-com bubble burst less than a year after that divergence appeared.
“The growing divergence that exists now and the outright negative hyperscalers price performance are reminiscent of the 1999-2000 dynamic,” Hunter wrote.
He is not saying AI is fake. He is not saying the market is about to collapse. What he is saying is that a crowded investment theme showing this kind of internal fracture has historically been dangerous.
AI stock market concentration risk
Hunter watched what happened in June and July when money started moving out of AI names.
It did not look like conviction. It looked like people quietly heading for the exit. No clear sector story. No broad rotation into cyclicals that would signal investors still believed in growth. Just position unwinds and a drift toward whatever felt defensive.
“The broadening and rotation isn’t like the 4Q25 occurrence, when the shift had a pronounced pro-cyclical theme,” he wrote. “This time around, it seems to be an unrelated combination of position unwinds, attempts to rotate away from crowded Technology exposure and a shift into portions of the market that could be construed as defensive in nature.”
More JPMorgan:
- JPMorgan CEO cuts to the chase on stock market danger
- JPMorgan is nearing $1 trillion for a reason investors missed
- J.P. Morgan’s stock price is flashing valuation warning
That matters because of how concentrated the AI trade has become. In the first half of 2026, roughly 87.5% of all venture dollars went into AI companies. Of that, 43% went to just two names: OpenAI and Anthropic.
At the same time, the four biggest AI spenders — Meta, Microsoft, Amazon and Alphabet — are on track to spend $725 billion on AI infrastructure this year, up 77% from last year’s record $410 billion, according to Tom’s Hardware. The Magnificent Seven ETF has pulled back about 7% from its 2026 peak.
When a trade gets that concentrated and the biggest spenders start underperforming the hardware suppliers, that is the pattern Hunter is flagging. Not the spending. The divergence.
Treasury yields, oil prices, and seasonal weakness add to stock market risks
Hunter is not only watching AI. He flagged several other risks that could amplify a selloff if the AI trade cracks.
Treasury yields have already been climbing. The 30-year hit a near two-decade high this month. Higher yields make future earnings worth less in today’s dollars, which hits growth stocks hardest.
Analysts who have spent 2026 pricing Nvidia, Microsoft, and the rest on long-duration earnings assumptions face a harder math when the discount rate goes up.
Oil is the second pressure point. Brent has gained 13% in the past two weeks and Morgan Stanley now sees it hitting $100 in Q4. If that happens, inflation expectations rise, the Fed’s options narrow and bond yields could move higher still.
Then there is the calendar. September has averaged a negative return for the S&P 500 every year going back to 1928, somewhere between 0.7% and 1% in the red.
It does not happen every year. But it happens enough that arriving at September with a fragile AI trade, climbing yields and rising oil is not a combination anyone should feel comfortable about.
Angela/Getty Images
What JPMorgan’s autumn stock market warning means for investors in 2026
JPMorgan is not calling the end of the bull market. Hunter’s overall posture remains constructive on U.S. stocks. His message is narrower than the headlines suggest.
The concern is that the market’s gains have been too dependent on a small group of AI-linked stocks, that the rotation away from those names has not been healthy, and that several external pressures, yields, oil, and the calendar, are arriving at the same time.
The S&P 500 and Nasdaq automatically concentrate in the largest names. An investor who has never consciously bought Nvidia or Microsoft may still have heavy exposure to both through passive funds.
The 2000 comparison should be read as a warning about concentration, not a prediction of collapse.
The companies at the center of today’s AI trade have real revenue, real profits and real cash flow. The dot-com companies mostly did not.
What they share with 2000 is the behavior of the market around them, not the fundamentals underneath.
Related: Warren Buffett sends investors 10-word stock market warning
