Chip stocks fell hard on October 15, 2026, as hotter-than-expected U.S. inflation data collided with growing investor doubts about how the AI buildout is actually being financed. Nvidia, AMD, Intel, Micron and SanDisk all traded lower, South Korea’s KOSPI dropped sharply overnight, and the sector capped a week that erased more than $1 trillion in AI-linked market value, according to CNBC.
The selloff is not a single-company story. It stretches from Santa Clara to Seoul, and it marks a shift in tone from a year defined almost entirely by AI optimism to one where investors are suddenly asking harder questions about debt, capital costs and return on investment. Here’s what happened, why it happened, and what it means for the chip industry heading into 2027.
What Triggered the October 15 Chip Stock Selloff
The immediate spark was fresh U.S. inflation data released ahead of the Wednesday session. The numbers pointed toward sticky price pressure, which investors read as a signal that the Federal Reserve could keep interest rates higher for longer, according to Investing.com. Higher rates matter enormously to chipmakers right now because so much of the AI infrastructure boom has been financed with borrowed money rather than cash on hand.
That inflation shock landed on top of an already nervous market. Investors had spent the prior session digesting warnings from two of the industry’s most influential executives about financing constraints, and a report that one of the biggest AI spenders might start reselling its own compute capacity. Put those pieces together and you get a selloff that hit nearly every major chip name at once, rather than a single stock stumbling on its own earnings miss.
Nvidia, AMD, Intel and Micron Lead the Declines
Nvidia opened the U.S. session down 2.63% at $215.49, after falling 1.8% in premarket trading, according to data cited by NDTV Profit. AMD fell further, dropping 4.44% to $545.47 shortly after the opening bell, building on a 3.5% premarket slide. Micron, which sits at the center of the memory-pricing story feeding into this whole episode, opened 4.57% lower at $872.34.
Intel’s decline was sharper still. The stock fell more than 3% in premarket trading, and one report put the broader intraday session drop at 5.9%. Research firm ROIC.ai, tracking the same session, recorded Nvidia down 3%, AMD down 5%, Intel down 5.9% and chip-equipment maker Marvell down 5.5%.
Premarket Moves vs. the Opening Bell
Worth noting: the premarket percentages and the opening-bell percentages don’t always match exactly, because premarket volume is thin and prices can still swing once full trading begins. That’s why AMD’s premarket move (3.5%) looks smaller than its opening print (4.44%) in the same dataset. The direction, though, was consistent across every reading: down, and down broadly.
| Company | October 15 Move | Detail | Source |
|---|---|---|---|
| Nvidia | -1.8% premarket / -2.63% at open | Opened at $215.49 | CNBC / NDTV Profit |
| AMD | -3.5% premarket / -4.44% at open | Opened at $545.47 | CNBC / NDTV Profit |
| Intel | -3%+ premarket / -5.9% intraday | Sharpest decline among large-caps | Investing.com / ROIC.ai |
| Micron | -4.4% premarket / -4.57% at open | Opened at $872.34 | CNBC / NDTV Profit |
| SanDisk | -7.27% at open | Fell to $1,301.33 | NDTV Profit |
| TSMC | -1.81% | Asian trading session | CNBC |
| Marvell | -5.5% | Chip-equipment and networking exposure | ROIC.ai |
| Qualcomm | -8% | Prior session, October 14 | Startup Fortune |
SanDisk, Marvell and Broadcom Get Pulled Into the Rout
SanDisk took the single biggest hit among the names tracked on October 15, sliding 7.27% to $1,301.33 shortly after the market opened. The stock’s size of move stands out because SanDisk had been one of the clearest beneficiaries of rising NAND prices earlier this year, which makes its reversal a useful signal of just how fast sentiment turned.
Broadcom was swept into the same selloff, though the available reporting does not give an exact percentage move for the stock on October 15. Marvell, which supplies custom silicon and networking components used across AI data centers, fell 5.5%. None of these companies reported new negative news about their own businesses that day. The moves were almost entirely sentiment-driven, tied to the macro backdrop rather than anything specific to their product roadmaps.
The Philadelphia Semiconductor Index Drags the Nasdaq Into Correction
Zoom out from individual stocks and the index-level damage looks even worse. The Philadelphia Semiconductor Index, known as the SOX and widely watched as the benchmark for chip-sector health, fell 4.3% in the session preceding the October 15 rout, according to Startup Fortune. That drop helped drag the Nasdaq 100 into correction territory, a threshold investors watch closely because it signals a shift from a healthy pullback into something more structural.
The available reporting doesn’t give a precise October 15 figure for the S&P 500, but the pattern across every index tracking tech and semiconductor exposure pointed the same direction that week: down, and accelerating.
South Korea’s KOSPI Plunge Shows This Is a Global Story
The selloff wasn’t confined to U.S. markets. South Korea’s benchmark KOSPI index plunged 7% overnight, a move that intensified the global memory-chip selloff rather than following it. South Korea is home to two of the world’s three major DRAM and HBM suppliers, so a sharp move there functions almost like a leading indicator for how the memory side of the AI supply chain is being priced.
Put simply, this wasn’t a Wall Street mood swing that spilled overseas. Asian markets moved first and harder, which supports the idea that the underlying worry (financing costs for AI infrastructure, and what happens if that spending slows) is a global concern rather than a U.S.-specific one.
Broadcom and Intel CEOs Say Capital, Not Chips, Is the Real Constraint
Some of the most pointed commentary ahead of the selloff came directly from two chip CEOs. Broadcom chief executive Hock Tan put it bluntly: “The biggest constraint is capital,” he said, according to CNBC. That’s a notable statement from the head of a company that has spent the past two years benefiting from AI-driven custom-silicon demand.
Intel CEO Lip-Bu Tan echoed the point in the same CNBC report: “If there’s any challenge in the front, it’s capital and the cost of capital, and then the other part is, in the longer run, people want to see the ROI.” Coming from Intel, a company in the middle of its own capital-intensive foundry turnaround, that comment lands as both a market observation and something close to a warning about his own company’s spending plans.
Hock Tan’s Casino Comparison
Tan went further in separate remarks, criticizing how some investors are approaching AI stocks altogether. “A lot of investors have no clue what they are investing in,” he said, according to Investor’s Business Daily, comparing the behavior to casino gambling. It’s an unusually blunt line from a sitting CEO of a company that has itself ridden the AI wave to record results, and it’s part of why the comments traveled so widely once the selloff began.
Inflation Data Reignites Fear of More Fed Rate Hikes
Back on the macro side, the inflation print that kicked off the October 15 session mattered because it undercut a narrative the market had been leaning on for months: that rates were on a clear path down. Skyler Weinand, chief investment officer at Regan Capital, summed up the mood directly: “It’s clear as ever that inflation is stubborn and moving in the wrong direction,” he said, according to Swissinfo.
That matters specifically for chipmakers because so much of the current AI data center buildout, from GPU clusters to new fabs, is funded through debt rather than free cash flow. If borrowing costs stay elevated longer than planned, the math behind years of projected AI returns gets less favorable, and investors have started pricing that risk in real time rather than waiting for it to show up in quarterly earnings.
Meta’s AI Compute Resale Plan Adds Fuel to the Selloff
Layered on top of the rate worries was a report that Meta was considering building a business to resell excess AI compute capacity, a detail that revived doubts about whether hyperscalers’ AI spending is actually producing the returns they’ve promised investors. If a company with Meta’s scale is looking at ways to monetize spare capacity rather than fully absorbing it internally, that’s a meaningful signal about utilization rates across the broader AI infrastructure build.
Strategist Neil Wilson of Saxo tied the whole picture together: “Worries about the AI buildout, bubbly valuations, lower-than-expected earnings growth at leading US frontier labs, Chinese competition, inflation and rates sent investors running for the covers,” he said, according to CNN. That’s a long list of causes, and that’s rather the point: no single factor explains the move, which is part of why it hit so many names simultaneously.
A Brutal Week: More Than $1 Trillion in AI Value Erased
Add it all up and the week ending October 15 wiped out more than $1 trillion in AI-linked stock market capitalization, according to CNBC’s tracking of the broader rout across chipmakers, memory suppliers and AI infrastructure names. That figure spans far more than Nvidia and AMD. It captures the memory suppliers, the networking and custom-silicon players, and the Asian manufacturers that sit upstream of nearly every AI server built this year.
A trillion dollars sounds abstract until you compare it to the gains that got the sector there in the first place. Much of that value was added over the course of 2026 on the strength of AI capex announcements, new chip generations and cloud providers racing to lock up GPU supply. Losing that much in a week doesn’t erase the underlying demand for AI compute, but it does reset how aggressively the market is willing to price in future growth without more evidence of near-term returns.
How This Selloff Stacks Up Against Earlier 2026 Routs
This isn’t the first time chip stocks have had a rough week in 2026. Earlier in the year, a separate slowdown call from a prominent analyst briefly sank Nvidia while lifting defensive names like CrowdStrike, before the sector recovered within weeks. What’s different this time is the breadth: the October rout hit the full chain at once, from GPU makers to memory suppliers to networking silicon, and it was compounded by two consecutive brutal sessions rather than a single bad day.
October 14 vs. October 15: Two Days, One Rout
The October 15 declines actually built on an even sharper session the day before. On October 14, Micron fell 13.2%, Qualcomm dropped 8%, Nvidia slid nearly 4%, and the SOX index fell 4.3%, according to Startup Fortune’s tracking of that session. Seen together, the two days form a single, connected rout rather than two unrelated events, with the October 14 declines in memory and mobile-chip names setting the tone for the broader October 15 slide across GPUs, CPUs and foundry-adjacent names.
| Date | Index / Stock | Move | Context |
|---|---|---|---|
| Oct. 14 | SOX (Philadelphia Semiconductor Index) | -4.3% | Dragged Nasdaq 100 into correction territory |
| Oct. 14 | Micron | -13.2% | Deepest single-stock drop of the week |
| Oct. 14 | Qualcomm | -8% | Set up the broader Oct. 15 slide |
| Oct. 14 | Nvidia | ~-4% | Part of a broad tech retreat |
| Oct. 15 | KOSPI (South Korea) | -7% overnight | Signaled a global, not just U.S., rout |
| Week total | AI-linked market cap | >$1 trillion erased | Spans chipmakers, memory suppliers, AI infrastructure names |
Market Impact: What a Capital Crunch Means for Chip Roadmaps
The practical question for anyone tracking the hardware industry is whether this selloff changes actual product roadmaps, or whether it’s purely a stock-price story. The honest answer right now is that it’s too early to say, but the comments from Hock Tan and Lip-Bu Tan matter because they came from inside the industry, not from outside analysts. When the CEOs of Broadcom and Intel are both pointing to capital costs rather than manufacturing capacity as the binding constraint, that’s a different kind of warning than a bear-case research note.
If financing does get tighter, the most exposed plans are the ones that depend on continuous, large-scale debt issuance: new fab construction, rack-scale AI server deployments, and the multi-year GPU supply commitments that cloud providers have been signing throughout 2026. None of that spending stops overnight, but the pace at which new commitments get signed could slow if borrowing costs stay elevated into 2027.
Competitive Comparison: Nvidia vs. AMD vs. Intel vs. the Memory Makers
Not every company is exposed to this story the same way. Nvidia, despite posting the smallest percentage decline among the large-caps tracked on October 15, carries the largest absolute dollar exposure simply because of its size and its central role in AI infrastructure spending. AMD’s exposure is tied more closely to its own aggressive AI accelerator roadmap and the capital intensity of matching Nvidia’s supply commitments.
Intel’s situation is arguably the most sensitive of the three, given that its foundry turnaround already depends on sustained capital spending at a moment when, by its own CEO’s admission, the cost of that capital is rising. The memory makers (Micron, SanDisk, and by extension the South Korean suppliers behind the KOSPI move) sit in a different position entirely. Their stocks are reacting to a layered story that includes both financing worries and separate oversupply concerns tied to new DRAM and HBM capacity expected to come online by 2029.
Historical Context: From Dot-Com Caution to the 2026 AI Trade
Chip-sector selloffs tied to financing anxiety aren’t new, but the scale of today’s AI buildout makes direct historical comparisons tricky. The closest precedent in tone, if not in exact mechanics, is the kind of capital-discipline scrutiny that hit telecom and infrastructure spending after the dot-com era, when investors eventually stopped rewarding growth stories that depended on continuous external funding and started demanding visible returns instead.
The difference this time is speed. Where that earlier shift in investor sentiment played out over a year or more, the October 2026 version compressed into roughly 48 hours, moving from a CEO comment about capital costs to a trillion-dollar weekly swing in market value. That compression is itself a signal of how closely AI infrastructure stocks are now being watched for any sign of a financing crack.
What Happens Next: Five Predictions
- Expect continued volatility tied to upcoming inflation prints and Fed commentary, since inflation data was the direct trigger for the October 15 leg of this selloff.
- Chip CEOs are likely to lean harder into capital-discipline language on upcoming earnings calls, echoing Hock Tan and Lip-Bu Tan’s framing around cost of capital and ROI rather than just unit shipments.
- Memory stocks will likely stay the most volatile corner of the sector, since they’re pricing both near-term financing worries and a separate, longer-dated oversupply concern tied to new capacity arriving by 2029.
- Watch for other hyperscalers to follow Meta’s reported lead on monetizing excess AI compute, since doing so would ease some of the utilization and ROI questions investors are raising.
- If the rout deepens, expect sharper scrutiny of the debt structures behind large AI data center financing deals, an area that has drawn relatively little public attention compared with chip supply itself.
None of these are certainties. They’re reasonable extrapolations from what two chip CEOs, a market strategist and a fixed-income investor were all pointing at in the same 72-hour window. The one thing that does look fairly settled is that “is AI spending sustainable” has moved from a background debate into the question actively moving billions of dollars in chip-sector market value, in real time, this week.
Frequently Asked Questions
Why did chip stocks fall on October 15, 2026?
Fresh U.S. inflation data raised fears of further Federal Reserve rate hikes, which hit chip stocks especially hard because so much AI infrastructure spending is financed with debt. That shock landed on top of CEO warnings about capital costs and a report about Meta considering reselling excess AI compute.
Which chipmakers fell the most in this selloff?
SanDisk recorded the sharpest single-day drop among the names tracked, falling 7.27% at the open. Intel fell as much as 5.9% intraday by one measure, while AMD, Micron and Marvell all posted declines above 4%.
What is the SOX index and why does it matter?
The SOX, or Philadelphia Semiconductor Index, tracks the largest U.S.-listed chip companies and is widely used as a proxy for sector health. Its 4.3% drop on October 14 helped pull the broader Nasdaq 100 into correction territory.
Did the selloff affect markets outside the United States?
Yes. South Korea’s KOSPI index plunged 7% overnight, and Taiwan Semiconductor Manufacturing Co. fell 1.81% in Asian trading, showing the rout extended well beyond U.S. markets.
What did Broadcom and Intel’s CEOs say about AI spending?
Broadcom CEO Hock Tan said capital, not chip or power supply, is now the biggest constraint on AI expansion. Intel CEO Lip-Bu Tan said investors increasingly want to see return on investment rather than just continued capital spending.
How much value did the chip sector lose this week?
More than $1 trillion in AI-linked market capitalization was erased across chipmakers, memory suppliers and AI infrastructure names during the week that included the October 15 session, according to CNBC’s tracking of the rout.
Is this selloff a sign the AI boom is ending?
Not necessarily. The decline reflects renewed scrutiny of financing costs and return-on-investment timelines rather than a drop in demand for AI compute itself. Chip CEOs framed it as a capital-cost and investor-expectations issue, not a demand problem.
What should investors watch next?
Upcoming inflation data, Federal Reserve commentary on interest rates, and how hyperscalers address AI compute utilization and financing on their next earnings calls are the clearest near-term signals for whether this rout extends or stabilizes.