Nvidia, Micron, AMD Lead Tech Sell-Off as AI Trade Cools

You’d think the AI revolution was on life support. That the party’s over. That the trillion-dollar buildout of data centers and GPUs is suddenly hitting a wall.

But look closer: Nvidia dropped 5.2% on Monday. Micron slid 4.8%. AMD lost 3.9% — the worst single-day losses for the sector in over a year. The Nasdaq 100 shed 2.3%, and the VanEck Semiconductor ETF (SMH) fell 3.7%. It wasn’t a crash. It was a rotation. A nasty one.

The AI trade — the engine that’s powered nearly 40% of the S&P 500’s gains this year — is finally catching a cold. And investors are scrambling to figure out if this is just a cough or something worse.

The Numbers Don’t Lie

Here’s what happened: Nvidia’s market cap vaporized roughly $80 billion in a single session. Micron, which was already down 12% from its 52-week high, slipped further after a report suggested memory chip demand could soften in the back half of 2025. AMD got caught in the crossfire, falling alongside its peers even though it reported solid data-center GPU sales just two weeks ago.

“This is a textbook profit-taking event,” says Sarah Collins, senior equity strategist at Alpine Capital Advisors. “You’ve had a parabolic move in AI names since October. The valuations are pricing in perfection. Any whiff of a slowdown — even seasonal — and the algorithms start selling first, asking questions later.”

Collins is referring to a pattern we’ve seen before. In mid-2023, a similar rotation hit the “Magnificent Seven” before they roared back. But this time feels different. Interest rates are higher for longer. The Fed’s dot plot now shows just one cut in 2025, not the three the market was betting on. That’s a headwind for growth stocks — especially ones trading at 40x+ forward earnings.

We’ve seen AI bubble comparisons to 1999 before, but this sell-off is manifesting in real-time: volume spikes, VIX jumping to 18, and options skew flipping bearish for the first time since October.

Why the Rotation Happened

Three catalysts collided last week. First, a hotter-than-expected producer price index (PPI) report rekindled inflation fears. Second, the Bank of Japan signaled a potential rate hike — which scares yield-sensitive money that’s been borrowing yen to buy U.S. tech stocks. Third, and most critically, AI monetization questions are no longer theoretical.

“Investors are starting to ask: ‘Who’s actually making money from AI, beyond the chipmakers?'” says David Okoro, technology analyst at Vector Research Group. “Microsoft and Meta are spending billions on infrastructure, but their cloud revenue growth rates are decelerating. The payoff timeline keeps getting pushed out. That’s a problem for a market that priced in immediate returns.”

Okoro points to a Reuters report that flagged potential order cuts from hyperscalers in Q3. Nvidia’s H100 lead times have shrunk from 52 weeks to just 8. That’s great for supply chains. But it’s also a demand signal — and not the kind bulls want to see.

Micron, meanwhile, is caught in a different trap. Its high-bandwidth memory (HBM) chips are crucial for AI accelerators, but the company is also heavily exposed to the PC and smartphone markets, which remain weak. “Micron is the canary in the coal mine. If HBM orders dip, the whole memory sector corrects,” explains Okoro.

What This Means for Main Street

If you don’t own individual AI stocks, you might think this sell-off is irrelevant. You’d be wrong. Tech is 30% of the S&P 500. Tech growth drives 401(k)s, pension funds, and state budgets. A sustained pullback in AI names could ripple into retail, cloud software, and even defensive sectors like utilities — paradoxically, since utilities benefit from AI data center power demand.

“The irony is that the same AI trade that caused the sell-off is the same trade that lifted utilities like Alliant Energy,” says Collins. “We’re seeing a reversal of a reversal. Capital is fleeing high-beta tech and rotating into boring dividend plays. That’s classic risk-off behavior.”

For individual investors, the message is blunt: rebalance now or regret later. The average retail portfolio has 15-20% allocated to tech — double the historical norm. If you’ve ridden Nvidia from $200 to $900, taking some chips off the table isn’t cowardice. It’s math.

And then there’s the global dimension — the sell-off isn’t just American. Taiwan’s TSMC slid 3.1%, South Korea’s SK Hynix fell 4.2%, and Japan’s Tokyo Electron dropped 2.9%. The AI trade is a global momentum bet, and momentum works both ways.

The Bigger Picture

Let’s be clear: AI isn’t going away. Generative AI adoption is still in the single digits vs. long-term potential. But markets don’t price 10-year horizons — they price 6-12 month realities. And for the next two quarters, the narrative is shifting from ‘how much can AI earn?’ to ‘how much are we overspending on AI infrastructure?’

That’s not a new question. In the early 1990s, Cisco and Sun Microsystems surged on internet backbone spending — only to crash when reality failed to justify valuations. The internet still changed the world. But the stocks that survived weren’t the ones priced for perfection in 1994.

“The winners in AI will be the ones that can demonstrate ROI in the next 18 months — not just in 2030,” says Collins. “Nvidia and AMD are great companies, but they’re priced as if AI monetization is a done deal. It’s not. And the market is finally acknowledging that.”

So where does that leave us? For the next few weeks, expect volatility. The CBOE NDX Volatility Index (VXN) is already signaling more pain ahead. Dollar-cost averaging into big tech might work. But buying the dip on Day One — that’s a gamble even the pros are avoiding.

One thing’s for sure: the easy money in AI is behind us. From here, it’s a stock-picker’s market. And the first rule of stock-picking? Don’t fall in love.