J.P. Morgan is telling you to sell the top in AI. Will you listen or not?

The Philadelphia Semiconductor Index fell 11% in two days.

SanDisk dropped over 10%, Micron dropped over 10%, and SK Hynix plummeted 14.57% in one day.

As soon as the news came out that Meta was selling idle computing power, the market panicked directly — "AI computing power oversupply" caused the entire hardware chain to collapse instantly.

The crazier the rally in the first half of the year, the more painful the drop in these two days.

And right before this bloodbath, J.P. Morgan had just done something —

Raised its S&P 500 year-end target from 7,600 to 7,800.

Bullish on one hand, warning on the other.

What did it say?

"AI trading is extremely crowded, and low-quality AI concept stocks could trigger a flash crash."

Then it gave a solution: the barbell strategy.

On one end, hold high-quality growth stocks and real AI beneficiaries; on the other, allocate some low-volatility stable assets as a buffer.

Translation: I'm still bullish on AI, but those junk AI stocks in your hands — get rid of them quickly.

Here's the question: Is this a normal risk warning, or is it really hinting that the time for sector rotation has come?

My judgment is —

Both. And it's more serious than you think.

In the report, J.P. Morgan outlined two paths:

Path 1 (optimistic): Cloud service providers and AI model companies slowly turn technology into real money, profitability catches up, the entire AI industry chain profits together, and the profit pie grows larger.

Path 2 (pessimistic): Chip companies make too much profit, squeezing the profits of downstream customers. Cloud service providers are forced to cut capital expenditures, which in turn causes chip demand to collapse, and the entire AI chain crashes together.

J.P. Morgan itself said it leans towards Path 1.

But it also added — the market consensus expectation is: cloud service provider capital expenditure grows 100% in 2026, plummets to 22% in 2027, and drops to just 7% in 2028.

If this expectation materializes, AI trading will face a "more significant and more sustained correction."

I'm bullish long-term, but don't be stubborn short-term.

On the other side, Bank of America: for the week ending July 1, U.S. equity funds saw net outflows of $17.2 billion, the largest weekly redemption record since March.

U.S. equity funds have had net outflows for two consecutive weeks.

Bank of America's bull-bear indicator jumped from 9.1 to 9.5, and the "sell signal" has lasted for six weeks without being lifted.

Note — the money isn't rotating from AI to other sectors.

It's directly leaving the market.

Some money went to bonds, some went to the Japanese stock market.

"Rotation" or "retreat"?

As it stands now, it looks more like a retreat.

So what should you do?

Don't fool yourself. Those second- and third-tier AI concept stocks and speculative small-cap semiconductor stocks you hold are exactly the "extremely crowded" part that J.P. Morgan mentioned.

The "barbell strategy" sounds fancy, but in plain words it's just one sentence:

Throw away the junk, hold the core, and keep cash ready.

Wait for the panic to pass, wait for the real AI commercialization data to come out, then it's not too late to enter.

"In a bull market, everyone is a stock god; only when the tide goes out do you know who's swimming naked. J.P. Morgan handed you swimming trunks — will you put them on?"

#gStocks代币化股票上线 #非农爆冷打压加息预期 #ETH突破1700 $BTC $META $SNDK