Three months ago, semiconductor stocks were the best trade on Wall Street. The VanEck Semiconductor ETF (SMH) had just posted its strongest quarter on record — a staggering 71% gain from April through June 2026. Then the third quarter began, and the floor gave out.
By mid-July, chip stocks had officially entered a bear market, wiping out trillions in market value in a matter of weeks and reigniting the question every investor is now asking: was the AI rally a bubble all along — and is it finally bursting?
Here's what actually happened, why it happened, and what the smartest people on both sides of the debate are saying.
What happened: a record quarter, then a bear market
The Philadelphia Semiconductor Index (SOX) peaked near 14,655 in late June, then slid roughly 20% to about 11,674 by July 17 — the threshold that officially defines a bear market. Across the global chip sector, the reversal erased an estimated $3 trillion-plus in market value, one of the sharpest technology drawdowns of the year.
The damage was broad and brutal:
- Micron Technology, up nearly 197% year-to-date at its peak, dropped as much as 13% in a single session — roughly $138 billion in market value gone in one day.
- Intel slid around 21% from its recent high heading into its earnings report.
- AMD fell 7–8%, while equipment makers Applied Materials, Lam Research, and KLA each shed about 4% on some of the worst days.
- Even Nvidia, the undisputed leader of the AI trade, wasn't spared, falling 2.2% on July 17 alone.
The SMH ETF, which tracks the 25 largest U.S. chipmakers, lost more than 17% in the first three weeks of July and posted its worst week since April 2025.
The strangest part? The selling accelerated despite blockbuster earnings. Samsung reported second-quarter operating profit of roughly 89.4 trillion won (about $58 billion) — an increase of more than 1,800% year over year — and its stock fell about 7%. Taiwan Semiconductor (TSMC) beat on profit and revenue, then dropped more than 3% after guiding for higher capital spending than expected.
When great news gets sold, the story isn't about earnings. It's about expectations.
Why the chips crashed: five forces hit at once
This wasn't a single-headline crash. Five pressures converged on a sector that had been "priced for perfection."
1. Valuations ran too far ahead of reality. After a run like Micron's or SanDisk's, even a flawless business can't grow into its stock price fast enough. When shares already discount years of optimistic assumptions, the smallest crack in the narrative triggers an outsized repricing. 2. China changed the math. Chinese startup Moonshot AI unveiled a new model it claims performs on par with the leading systems from OpenAI and Anthropic. Days earlier, reports surfaced that DeepSeek — already one of the most compute-efficient AI labs in the world — is developing its own proprietary chip to reduce its dependence on Nvidia. Cheaper, open-source Chinese models raise an uncomfortable question: if AI gets radically more efficient, does the world really need to buy this many chips? 3. AI spending fatigue set in. As Edward Jones strategist Angelo Kourkafas put it, the market is starting to penalize companies that ramp spending too aggressively, while end-user demand for AI grows more price-sensitive. Investors want to see a return on the hundreds of billions in AI infrastructure — and the payoff timeline is getting harder to justify. 4. The Fed turned hawkish. Under Chair Kevin Warsh, the Federal Reserve has taken a firmer line on rates. Higher-for-longer rates are kryptonite for high-multiple growth stocks, whose value depends heavily on distant future earnings. 5. Profit-taking after a parabolic run. After the sector's roughly 65% climb in the first half of 2026, some of the selling was simply traders locking in extraordinary gains — mechanical, but painful.The bubble debate: dot-com echoes vs. "third inning"
This is where it gets interesting, because Wall Street is genuinely split.
The bears point to uncomfortable historical parallels. Bank of America strategist Michael Hartnett's proprietary "Bubble Risk Indicator" hit 0.91 — well above the Nasdaq 100's own reading — with levels of market concentration and overbought conditions he says haven't been seen since June 2000, just before the dot-com bubble burst. When a handful of names drive the entire index, the whole market becomes fragile. The bulls counter that this is a correction, not a collapse. Wedbush's Dan Ives called it "the third inning, one out, in a nine-inning game." FactSet data projects second-quarter 2026 semiconductor industry earnings will grow around 131% year over year. Morgan Stanley labeled the drop a "mid-cycle reset" rather than a market top. And on valuation, Goldman Sachs notes Nvidia's forward price-to-earnings ratio of about 21.7 is actually cheap relative to its five-year average near 72. Analyst 12-month price targets still imply substantial upside — roughly 56% for Nvidia and 66% for Micron from current levels.The key distinction: this selloff was driven by valuation and sentiment, not deteriorating fundamentals. Micron's revenue, Samsung's profits, and Nvidia's order book didn't get worse. What changed was how much investors were willing to pay for them.
Where the money went: the flight to "boring"
Capital rarely disappears — it rotates. As chips and AI names bled, investors sought shelter in exactly the kind of unglamorous, cash-generating business the market ignored for two years: Warren Buffett's Berkshire Hathaway.
Berkshire rose even as the broader market fell, helped by optimism that the conglomerate stepped up share repurchases — analysts estimate it bought back somewhere between $5 billion and $11 billion of its own stock in the second quarter. Its sprawling mix of insurance, energy, railroad, and industrial businesses suddenly looked like the safest room in the house. The rotation from high-multiple growth to durable value is one of the clearest signals of what this selloff really represents.
What to watch next
The next few weeks will decide whether this is a reset or the start of something worse. Key catalysts on the calendar:
- Intel earnings (July 23) — a struggling chipmaker's guidance will test whether the sector can stabilize.
- Samsung's full quarterly report (July 30) — a disappointment could reignite selling across Asian and U.S. chip names.
- Forward guidance from Nvidia, AMD, and Micron — if hyperscaler AI spending keeps outpacing revenue growth without a clear payoff, investor patience could run thin.
- The Fed and inflation data — any further hawkish surprise from Warsh would keep pressure on growth stocks.
Add in geopolitical noise — oil prices have climbed amid renewed U.S.–Iran military exchanges — and volatility is likely to stay elevated.
The bottom line
The July 2026 chip crash was real, fast, and historically sharp. But "bear market" and "bubble bursting" are not the same thing. The evidence so far points to a crowded trade unwinding under the weight of its own valuations — not a collapse in the underlying AI story. The memory shortage is real. AI infrastructure demand is real. What wasn't sustainable was expecting these stocks to rise in a straight line forever.
Whether this is the "third inning" or "June 2000 all over again" won't be clear for months. For now, the smartest posture is the one the market just relearned the hard way: respect valuations, question the narrative when everyone agrees, and never confuse a great company with a great price.
Frequently Asked Questions
Is the AI bubble bursting in 2026?
Semiconductor stocks entered a bear market in July 2026, but most analysts describe it as a valuation-driven correction rather than a fundamental collapse. Chip company earnings and AI demand remained strong; what changed was investor willingness to pay elevated prices. Whether it becomes a full bubble burst depends on upcoming earnings guidance and AI-spending returns.
Why did semiconductor stocks crash in July 2026?
Five factors converged: stretched valuations after a 65%+ first-half rally, new competition from Chinese AI models (Moonshot and DeepSeek), fatigue over massive AI capital spending, a hawkish Federal Reserve under Kevin Warsh, and profit-taking. The Philadelphia Semiconductor Index fell over 20% from its June record.
How much did the AI chip selloff wipe out?
Estimates put the global semiconductor market-value losses at more than $3 trillion from the June 2026 peak. Individual names like Micron lost roughly $138 billion in a single session.
Are chip stocks a buy after the 2026 selloff?
Analysts are divided. Bulls note that forward valuations (Nvidia's forward P/E near 21.7 vs. a 72 five-year average) and projected earnings growth of ~131% look attractive, with 12-month targets implying significant upside. Bears warn of dot-com-style concentration risk. This is not investment advice — do your own research and consider your risk tolerance.
Where are investors moving money during the AI selloff?
Many rotated into defensive value plays. Berkshire Hathaway rose during the selloff as investors sought its diversified, cash-generating businesses as a safe haven from volatile AI and semiconductor names.
---
This article is for informational and educational purposes only and does not constitute financial, investment, or trading advice. All market data reflects the period through July 21, 2026, and is subject to change. Always consult a licensed financial professional before making investment decisions.
