Chip stocks had their worst trading day in more than six years on Friday, June 5, 2026. The Philadelphia Semiconductor Index, known by its ticker SOX, fell 10.3% to close at 12,220.76, down from Thursday’s close of 13,617.50. It was the index’s steepest single-session drop since March 2020, and every one of its 30 components finished lower. U.S.-listed chipmakers lost close to $1.3 trillion in combined market value in that one session, with Nvidia, Micron, AMD, and Broadcom absorbing the deepest cuts.
The strange part is who got blamed. Broadcom, the company most investors pointed to as the trigger, had just reported one of the best quarters in its history. Revenue hit a record $22.19 billion. AI semiconductor sales grew 143% year over year. Broadcom even raised its full-year AI chip revenue forecast to $56 billion and reaffirmed a target of more than $100 billion for fiscal 2027. None of that stopped the stock from falling almost 20% over two sessions, and it didn’t stop the rest of the sector from falling alongside it. Four days later, Wall Street is still arguing about whether June 5 was a healthy reset after an overheated run or the first real crack in the AI infrastructure trade.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Happened: Chip Stocks’ Worst Day Since March 2020
The sell-off built over the course of the week, then accelerated fast. Broadcom’s earnings landed Wednesday evening, June 3. By Friday, a stronger-than-expected May jobs report had reignited fears that the Federal Reserve would delay rate cuts, sending Treasury yields higher and pulling money out of the most expensive corners of the stock market. Chip stocks, priced for a flawless AI buildout, were the most expensive corner of all.
“U.S.-traded chipmakers plunged on Friday, losing about $1.3 trillion in market value, with deep losses in AI heavy hitters including Nvidia, Micron Technology and Advanced Micro Devices.”
Reuters, market report, June 5, 2026, via Reuters
The damage spread well beyond the chip sector by the closing bell. The Dow Jones Industrial Average dropped roughly 680 points, and the Nasdaq Composite posted its worst single-day decline in more than a year, down 4.18%. The SOX index alone accounted for a disproportionate share of that pain, since semiconductor names had grown into some of the largest weightings in major indexes during 2026’s AI rally.
Broadcom’s Earnings Beat That Wasn’t Enough
Start with what Broadcom actually reported, because the numbers make the stock’s reaction look almost irrational on the surface. Fiscal second-quarter revenue reached $22.19 billion, up 48% year over year and a company record. AI semiconductor revenue, the number Wall Street cares about most, hit $10.8 billion, up 143% from a year earlier. Broadcom guided third-quarter revenue to about $29.4 billion, ahead of the roughly $28.5 billion analysts had modeled, and told investors it now expects $56 billion in AI semiconductor revenue for the full 2026 fiscal year. It also reaffirmed guidance for AI chip revenue to top $100 billion in fiscal 2027, a number that would sound absurd coming from almost any other chip company. Full details are in our earnings breakdown from earlier in June.
Despite the beat and raise, Broadcom shares fell 13.4% in the session immediately after the report, then another 7.9% on June 5 as the broader market rout hit, for a combined two-day decline of almost 20%, according to CNBC’s earnings coverage. The stock had run up so far and so fast that “reaffirm” read to some investors like a warning sign instead of a vote of confidence. Analysts had been modeling even steeper acceleration in custom AI chip demand, and when Broadcom’s forward commentary landed closer to its prior guidance than to the boldest Street estimates, the stock got treated like it had missed.
Which Chip Stocks Got Hit Hardest
Not every chip stock fell by the same amount, and the pattern says something about how investors were sorting winners from losers even during a broad panic.
Custom Silicon Names Led the Declines
Marvell Technology, which designs custom AI accelerators for hyperscale customers much like Broadcom does, fell 16.7%, the worst single-day move among the 30 SOX components. Micron dropped 13.2%, erasing roughly $150 billion in market value, as investors reassessed how much of the memory maker’s rally was tied to the same AI capex assumptions now in question. Arm Holdings lost 12.8%, and Intel slid 11.3%.
Merchant Chipmakers Held Up Relatively Better
Nvidia, the largest company in the sector by market value, fell 6%, wiping out more than $300 billion in market capitalization on its own but still losing less, in percentage terms, than most of its peers. AMD dropped almost 11%. ON Semiconductor fell 11.1% and Qualcomm slid 11%. Investors treated Nvidia’s broader, more diversified customer base as a relative safe harbor inside an otherwise ugly session, even as its dollar losses dwarfed everyone else’s.
June 5, 2026 Chip Stock Selloff, Company by Company
The table below breaks down how the session hit each major name, based on closing prices reported that day.
| Company / Index | June 5 Move | Two-Day Move | Approx. Market Cap Impact |
|---|---|---|---|
| SOX Index (broad) | -10.3% | ~-12% | ~$1.3T sector-wide |
| Marvell (MRVL) | -16.7% | – | – |
| Micron (MU) | -13.2% | – | ~$150B erased |
| Arm Holdings (ARM) | -12.8% | – | – |
| Intel (INTC) | -11.3% | – | – |
| ON Semiconductor (ON) | -11.1% | – | – |
| Qualcomm (QCOM) | -11.0% | – | – |
| AMD (AMD) | -10.9% | – | – |
| Broadcom (AVGO) | -7.9% | ~-20% | – |
| Nvidia (NVDA) | -6.0% | – | >$300B erased |
Source: company closing prices as reported June 5, 2026, and Reuters sector coverage.
The Real Trigger: A Hot Jobs Report Meets an Overheated Trade
Broadcom’s earnings reaction explains why chip stocks were already wobbly heading into Friday. It doesn’t fully explain why the whole sector fell off a cliff on June 5 specifically. That part traces back to the May jobs report, released the same morning, which showed the U.S. economy adding roughly 172,000 jobs, well above what markets expected. A labor market that strong gave the Federal Reserve less reason to cut interest rates soon, and rate-sensitive, high-multiple growth stocks sold off across the board as Treasury yields climbed.
Chip stocks, which had become some of the most richly valued names in the market after 2026’s run-up, were the obvious place for that repricing to land hardest. The same sector had added roughly $2 trillion in value earlier in the year, and stocks that climb the fastest tend to fall the hardest when sentiment turns, even briefly. Combine a hawkish jobs number with a Broadcom reaction that read as “as good as it gets,” and the entire AI-chip trade got repriced in a single session.
Wall Street’s Rocky Attempt to Stabilize
Monday, June 8, looked like relief at first. The SOX index jumped 5.6%, and Nvidia, Broadcom, and Micron all rose between 1.5% and 3.9% in premarket trading. But the bounce didn’t hold. By the closing bell, the S&P 500 and Nasdaq both finished lower, and Micron gave back most of an earlier intraday gain that had approached 10%, closing down about 1%. Broadcom slipped roughly 1% as its own rebound fizzled. The Motley Fool’s coverage captured the mood well: nobody was confident the worst was over.
Tuesday, June 9, brought more of the same uneven trading. The Nasdaq gained 0.86% and the S&P 500 added 0.30%, a modest recovery on paper. But the session came with its own scare: Super Micro Computer announced a surprise $7 billion equity raise to fund AI server orders, and the stock fell 7.6% on the news. Arm Holdings and ServiceNow each dropped more than 6% as investors flagged renewed geopolitical risk on top of everything else. Four trading days after the initial crash, chip stocks were stabilizing in the sense that they had stopped falling every session, not in the sense that anyone had resolved the underlying argument.
Why the Rebound Didn’t Settle the Argument
Part of what makes this crash hard to read is that the underlying chip market kept growing the entire time. The Semiconductor Industry Association reported that global chip sales hit $110.5 billion in April 2026, up 11% from March and up 93.9% from April 2025. The trade group projects full-year 2026 sales will top $1.5 trillion globally, which would make it the industry’s biggest year on record by a wide margin, according to the Semiconductor Industry Association.
That’s the tension driving every argument on trading desks right now. The physical chip business is not slowing down by any measure investors can currently see. What sold off on June 5 wasn’t demand. It was the price investors were willing to pay for that demand, after a run so steep that even a 143% jump in AI revenue wasn’t enough to keep expectations satisfied.
The AI Capex Question Hanging Over Every Chip Stock
Every chip stock in the SOX index is ultimately a bet on how much money hyperscalers and cloud providers keep spending on AI infrastructure. Broadcom’s own numbers argue that spending is still accelerating, not slowing. A 143% jump in AI semiconductor revenue, delivered in the same quarter as record margins, is not what a spending slowdown looks like on a balance sheet.
But equity investors don’t only price today’s revenue. They price the next several years of it, and June 5 showed how quickly that pricing can snap back when the pace of good news merely matches expectations instead of beating them by an ever-widening margin. Deloitte’s 2026 semiconductor industry outlook frames this as the central risk facing the sector this year: valuations built on the assumption that AI infrastructure spending keeps accelerating indefinitely, in a market where no hyperscaler has actually promised that. Whether cloud providers keep spending at the rate Wall Street built into chip valuations is something investors will only be able to judge one earnings season at a time, as each major provider reports results through the rest of 2026.
How This Crash Stacks Up Against Past Semiconductor Downturns
Semiconductor stocks have a long history of booming and busting harder than the broader market, and June 5 is far from the sector’s worst day on record. It’s severe by one-day standards. It’s mild compared to the multi-year downturns that have hit the industry roughly once a decade.
| Downturn | Approx. Peak-to-Trough Decline | Primary Trigger |
|---|---|---|
| 2000-2002 dot-com bust | About -80% over roughly 2.5 years | Internet and telecom capex collapse |
| 2007-2008 financial crisis | About -60% over about 16 months | Global credit crisis |
| 2011 European debt crisis | About -30% over about 8 months | Eurozone crisis, Japan and Thailand supply shocks |
| 2018 U.S.-China trade war | About -25% over about 9 months | Tariffs, memory price downcycle |
| 2020 COVID crash | About -35% in roughly 4 weeks | Pandemic shutdown shock (fastest recovery on record) |
| 2021-2022 rate-driven correction | About -45% over about 10 months | Inflation, rate hikes, PC and memory bust |
| June 2026 crash | -10.3% in a single session | Broadcom guidance reaction plus a hawkish jobs report |
Reuters, describing the scale of the move, called it the index’s largest one-day percentage plunge since March 2020, which itself was a four-week crash rather than a single session. That comparison is useful context. A steep one-day drop and a sustained bear market are different animals, and so far, June 5 looks much more like the former.
Market Impact: Nvidia, AMD, Intel and the Broader AI Trade
The ripple effects extend well past the chip sector itself. Memory suppliers have already pushed DRAM and NAND prices up sharply in 2026, and a sustained slowdown in AI infrastructure orders would eventually show up in that supply chain too, though nothing in the current data suggests memory demand is cooling. Data center landlords, power utilities serving AI campuses, and the software companies selling into hyperscaler budgets all have a stake in whether June 5 turns out to be noise or signal.
For everyday buyers, the more immediate effect is psychological rather than financial. GPU and CPU pricing hasn’t moved because of a single bad trading session, and it won’t. But the crash is a reminder of how tightly gaming hardware pricing, data center chip demand, and AI infrastructure spending have become linked since 2025. When AMD and Nvidia compete for the same hyperscaler AI budgets, swings in that budget eventually touch every product line each company sells, gaming GPUs included.
The Competitive Landscape: Custom Silicon vs. Merchant GPUs
June 5 exposed a real difference in how the market treats the two dominant business models in AI chips. Broadcom and Marvell build custom AI accelerators for a small number of enormous customers, mostly hyperscalers designing their own silicon. That model produces spectacular growth rates, like Broadcom’s 143%, but it also concentrates risk in a handful of contracts. Any hint that one customer might delay a ramp or renegotiate volume can move the stock hard in either direction.
Nvidia and AMD sell merchant GPUs to a much wider customer base, from hyperscalers to enterprises to research labs, which is part of why Nvidia’s percentage decline on June 5 was smaller than Broadcom’s or Marvell’s despite its far larger absolute dollar loss. Nearly all of these chips, regardless of who designs them, still get manufactured at TSMC’s advanced fabs, which means the foundry sits at the center of the story no matter which side of the custom-versus-merchant debate wins out.
What Analysts and Reporters Are Saying
Wire coverage on the day of the crash focused on how unusual it was to see a beat-and-raise quarter coincide with a sector-wide rout.
“The Philadelphia SE Semiconductor Index suffered its largest one-day percentage plunge since March 2020, erasing more than $1 trillion in stock market value.”
Reuters, market wrap, June 5, 2026, via Reuters
A separate Reuters dispatch the same day put the magnitude in blunt terms, describing how “the PHLX chip index slumped 10.3% in its deepest one-day loss since March 2020.” That framing, repeated across financial media through the following days, is why the crash immediately drew comparisons to the pandemic crash rather than to a routine pullback. Coverage of the June 8 attempted rebound carried a more cautious tone, with reporters noting that early gains failed to hold through the close, a pattern that tends to worry traders more than a clean one-day drop followed by a clean recovery.
5 Predictions for Chip Stocks Through the Rest of 2026
1. Volatility becomes the norm around every hyperscaler earnings report. Expect chip stocks to swing hard around Microsoft, Amazon, Google, and Meta results for the rest of the year, since each one now functions as a referendum on AI capex plans.
2. Custom silicon stays more volatile than merchant GPUs. Broadcom and Marvell’s customer-concentration risk means their stocks will likely keep swinging harder than Nvidia’s in both directions, purely because a handful of contracts can move the growth rate so much.
3. Memory makers stay relatively insulated. Micron’s business is underpinned by genuine supply tightness in DRAM and HBM, not just an AI growth narrative, which should make it less exposed to sentiment-driven swings than pure AI accelerator plays.
4. Macro data outweighs company fundamentals in the short term. As June 5 showed, a single jobs report can do more damage to chip stocks in an afternoon than a disappointing product cycle could do in a quarter. Expect Fed policy signals to keep driving single-day swings through the rest of 2026.
5. More double-digit single-day moves are likely. With valuations still elevated across the sector even after June 5, any hyperscaler that signals slower capex growth, even a small deceleration, could trigger another sharp, sector-wide repricing before the year is out.
Frequently Asked Questions
What caused the chip stocks crash on June 5, 2026?
A combination of two things hitting at once: a stronger-than-expected May jobs report that revived fears the Federal Reserve would delay rate cuts, and lingering disappointment over Broadcom’s earnings guidance from two days earlier, which investors read as a sign the AI chip upgrade cycle might not accelerate as fast as priced in.
How much money did the semiconductor selloff wipe out?
Close to $1.3 trillion in combined market value across U.S.-listed chipmakers in the single session on June 5, 2026, according to Reuters. Nvidia alone lost more than $300 billion in market capitalization, and Micron lost roughly $150 billion.
Did Broadcom actually miss earnings expectations?
No. Broadcom beat on revenue, posting a record $22.19 billion, up 48% year over year, and grew AI semiconductor revenue 143%. It also raised its full-year AI chip revenue guidance. The stock fell anyway because expectations had climbed even higher than the beat.
Which chip stocks fell the most on June 5?
Marvell Technology led the declines, down 16.7%, followed by Micron down 13.2%, Arm Holdings down 12.8%, and Intel down 11.3%. Nvidia’s 6% decline was among the smallest in percentage terms, despite its dollar losses being the largest of any chip stock.
Did chip stocks recover after the crash?
Partially. The SOX index rose 5.6% on June 8, but the broader market gave back those gains by the close, and several names including Micron and Broadcom pared their bounces. Trading through June 9 remained choppy rather than a clean recovery.
Is this the start of an AI stock bubble bursting?
There’s no consensus yet. Underlying chip demand kept growing through the crash, with the Semiconductor Industry Association reporting a 93.9% year-over-year jump in April 2026 sales. What sold off was valuation, not demonstrated demand, though that distinction may not hold if hyperscaler spending actually slows later in the year.
How does this compare to the 2020 COVID chip stock crash?
It’s smaller so far. The 2020 crash unfolded over about four weeks and cut semiconductor stocks by roughly a third. The June 2026 crash delivered a steeper single-day move but, as of this writing, hasn’t turned into a sustained multi-week decline.
