$1.3 trillion vanished Friday. Bubble, or just profit-taking? June 7th 2026

Curated by Alexis

AI and chip stocks shed roughly $1.3 trillion on Friday, the semiconductor sector's worst day since 2020, after a hot jobs report spiked interest-rate fears and Broadcom's outlook rattled the chip trade. The sharpest people in finance flatly disagree on what it means: the bubble finally cracking, or profit-taking after a euphoric run. Here is the case for each, with the receipts. You decide.

Quick Hits

The rout — what broke on Friday

Stocks had their worst day in over a year. The Nasdaq fell 4.2% and the S&P 500 dropped 2.6%, the worst session since April 2025, as AI names tumbled and the odds of a Federal Reserve rate hike rose on a stronger-than-expected jobs report.
Semiconductors led the carnage. Chip stocks slid hard after Broadcom's AI-chip outlook disappointed: Nvidia fell about 6% and dipped below a $5 trillion valuation, with Micron, AMD, and Marvell falling alongside it.

The case for "just profit-taking"

The Dow hit a record the same day. Even as chips cratered, the Dow climbed to a record high as money rotated into health care and financials. That is the signature of a sector rotation, not a market-wide flight from stocks.
The classic bubble-burst signals are missing. Wall Street analysts point out that corporate earnings never collapsed and there has been no dot-com-style IPO frenzy, leading some to argue the AI opportunity is still early rather than ending.
Goldman's CEO says the AI selloffs are "too broad." David Solomon has argued the rout is overdone, expecting AI to produce "winners and losers" and "plenty of companies" to "pivot and do just fine" rather than face wholesale destruction.

The case for "the bubble is bursting"

Ray Dalio says it is a bubble that will burst. The Bridgewater founder warned that the AI market is showing the classic signs of a bubble that will eventually pop as paper wealth gets converted back into cash.
BofA says the chart looks like March 2000. Bank of America's Michael Hartnett told clients the market just echoed the dot-com top, with gains dangerously concentrated in a sliver of stocks, and pushed his closely watched Bull & Bear indicator into "sell" territory.
The math still does not close. By Sequoia's widely cited estimate, the AI industry needs to earn roughly $600 billion a year to justify its hardware spending — a shortfall that has intensified fears of an AI bubble.

How you'll know who's right

Both camps are credible, and Friday did not settle it. Start with what actually triggered the drop: a strong jobs report that pushed up bond yields and the odds of a rate hike. That is a macro scare, not an AI-demand miss. It hit the most crowded trade in the market hardest, which is exactly what you would expect whether or not there is a bubble underneath.

So the honest test is structural, and it comes down to one gap. Hyperscalers are committing on the order of $700 billion to AI infrastructure this year, while the applications built on top still earn a small fraction of that. The bull case says the gap closes as cloud-AI revenue compounds, and the run-rates at AWS, Azure, and Google Cloud are in fact growing fast. The bear case says the gap is too wide and widening, and that the bill eventually arrives as writedowns.

Three things will tell you which way it breaks before the pundits agree: the Fed's rate path from here, the next round of hyperscaler capex guidance, and whether the market's gains broaden out or stay stuck in a handful of names. Watch those, not the one-day moves.

Key Takeaways

The trigger was macro, not AI-specific. A hot jobs report and a rate-hike scare hit the market's most crowded trade. Friday said as much about bonds as about AI.
The bull tell: earnings have not broken and the Dow set a record the same day. That reads more like rotation than capitulation.
The bear tell: the capex-versus-revenue gap is real and widening, and serious investors are now naming a bubble out loud.
The swing factor: rates and the next capex guidance. If hyperscalers blink on spending, the bear case gets its proof.