TLDR;
On June 5, the global chip sector shed over $1.3 trillion in market cap in a single session.
The trigger: Broadcom grew AI revenue 143% year-over-year to $10.8 billion — and the market punished it anyway.
Why? Next-quarter guidance came in at $16B against a $17.2B expectation, and the full-year outlook wasn’t raised.
Nvidia briefly lost its $5 trillion valuation crown; AMD and Intel each fell roughly 11%.
The lesson: the AI trade isn’t priced for growth. It’s priced for perfection — and perfection now means beating an impossible number.
Let me show you the most counterintuitive number in business this month.
A company grew its AI revenue 143% in a year. Not 14%. Not 43%. One hundred and forty-three percent, to $10.8 billion.
And the market’s response was to wipe it — and everything near it — by over a trillion dollars in an afternoon.
That company was Broadcom, and the date was June 5. The damage didn’t stay contained, either. Nvidia briefly gave up its $5 trillion valuation crown. AMD and Intel each fell around 11%. More than $1.3 trillion in combined market cap evaporated in one session — for a sector that, by every fundamental measure, is having the best year of its existence.
So what happened?
Growth wasn’t the problem. The expectation was.
Here’s the math that actually moved the market. Broadcom guided next quarter to roughly $16 billion in AI revenue. The expectation was $17.2 billion. And it declined to raise its full-year forecast.
That’s it. That’s the whole crime. Grow 143%, miss a guess by 7%, and the market treats you like you’ve failed.
What Broadcom deliveredWhat the market wanted+143% AI revenue growthMore$10.8B AI revenueMore~$16B next-quarter guide$17.2BFull-year outlook heldFull-year outlook raised
This is what “priced for perfection” looks like in practice. When a valuation already assumes the best possible future, merely delivering an excellent one is a letdown. The number to beat isn’t last year’s result. It’s the fantasy baked into the price.
The thing investors keep forgetting
There’s a difference between a great business and a guaranteed one, and crowded valuations quietly erase that distinction.
The AI infrastructure story is real. Chips genuinely sit at the foundation of every data center being built. But “real demand” and “demand that grows fast enough to justify this valuation, every single quarter, forever” are two completely different claims — and the market had stopped pricing the difference.
June 5 was the difference reasserting itself.
What I’d take from it
Not “the AI trade is over.” That’s the lazy read, and it’s probably wrong — the underlying buildout is enormous and still early.
The real takeaway is narrower and more useful: when a valuation leaves no room for anything short of perfection, the quality of the news stops mattering. The only thing that moves the price is the gap between reality and expectation — and when expectations are set in fantasy, even spectacular reality disappoints.
143% growth got punished. Remember that the next time someone tells you a number can’t go down because the company is “too good.” Good was never the question. Priced-for-perfect was.