Chip Stocks' 8-Day Losing Streak Is the Longest Since 2006 - Short Sellers Made $32 Billion, But BofA Says Valuations Already Match Post-Dot-Com Cisco

What Happened

Chip stocks fell again on Friday, October 16, pushing the Philadelphia Semiconductor Index (SOX) to an eighth straight losing session. According to Bloomberg, that's the index's longest losing streak since May 2006 - roughly two decades. Over that stretch the index shed somewhere around 12-13% of its value, wiping out an estimated $2 trillion in sector-wide market capitalization. And according to S3 Partners data cited by Bloomberg, short sellers who bet against these stocks during the slide are sitting on roughly $32 billion in paper profits.

The broader market wasn't spared either. The S&P 500 fell 84.62 points (1.1%) to close at 7,659.44, while the Nasdaq Composite dropped a sharper 572.99 points (2.1%) to 26,140.03, per AP's market wrap. The Dow Jones Industrial Average held up relatively better, falling 175.28 points (0.3%) to 52,002.41 - a reflection of its lighter weighting in AI and chip names and heavier tilt toward energy and old-economy industrials. Small-caps actually fared worse than the mega-cap-heavy Nasdaq: the Russell 2000 dropped 1.7%. Among individual names, Intel slid 5.8%, and memory giant Micron fell roughly 13% for the week, sliding from its October 7 all-time closing high of $1,088 down to around $894. CoreWeave, still working through the aftershocks of its mid-October bond-market scare (its credit-default-swap spread spiked to distress levels on October 14-15), was down roughly 25% for the week, keeping anxiety about AI infrastructure financing in the headlines.

Against that backdrop, a research note from Bank of America semiconductor analyst Vivek Arya drew particular attention. As reported by Benzinga, Arya pointed out that Nvidia, Broadcom and Micron are now trading around 13 times forward earnings - squarely inside the 10-to-20-times range Cisco Systems traded at in the years after the dot-com crash. His framing of the irony: "Investors are paying post-crash multiples for a cycle that hasn't crashed."

Why Valuations Are Already Depressed Even Though Earnings Haven't Cracked

What makes Arya's note worth sitting with is that it highlights exactly how this selloff differs from 2000. Cisco, briefly the most valuable company on earth in March 2000, peaked near $82 a share before collapsing almost 90% to $8.12 by October 2002. That crash was fundamentally a demand problem: telecom and internet companies had over-ordered networking gear during the bubble, and when the bubble popped, the demand itself evaporated. Cisco's actual revenue and earnings fell hard, and the stock didn't reclaim its March 2000 peak until March 2026 - a 26-year round trip.

Nvidia, Broadcom and Micron haven't seen anything like that yet. Arya's note argues that AI chip utilization remains near full capacity, and rental prices for even older-generation Nvidia GPUs are still rising - a very different signal from a 2000-style demand collapse. What's happening instead is a valuation reset, not an earnings reset: investors, worried about a future AI bust, have already priced these stocks as if that bust happened, regardless of whether the underlying business cycle has actually turned. That's a psychological and positioning phenomenon that can show up in the share price well before - or even without - any change in the fundamentals.

The broader market's reaction tells the same story from a different angle. The fact that short sellers have banked $32 billion during an eight-day skid means a large pool of capital is actively betting the decline continues. Meanwhile, the Dow's relative resilience and the Russell 2000's outsized drop both point to a selective unwind - money leaving one crowded trade (AI and chips) rather than a broad market panic. It's also worth noting that JPMorgan, in a note roughly a month earlier, had argued the AI trade's crowded positioning was already clearing out and turned constructive on semiconductors; the fact that the losing streak has since stretched into a 20-year record suggests that call is still being tested by the tape.

What to Take Away From This

• Valuation compression and earnings collapse are not the same thing. A falling stock price alone doesn't tell you whether the underlying business is deteriorating. When revenue and profit forecasts are holding steady while the price-to-earnings multiple gets crushed by fear, the eventual recovery path can look very different from a scenario where earnings themselves are cratering.
• Short-seller profits don't guarantee the trend continues. A $32 billion paper gain confirms the selloff has been severe, not that it will keep going. Crowded short positions can also set up sharp short-covering rallies if sentiment turns even slightly.
• Historical analogies are reference points, not forecasts. Cisco's 26-year round trip is a useful cautionary extreme, but investors should weigh for themselves whether the defining condition behind that crash - an actual evaporation of demand - applies here.
• Watch the dispersion, not just the headline index. The Dow holding up while the Russell 2000 and Nasdaq fell harder is a sign this move is concentrated in a specific theme (AI/semis) being unwound, rather than a market-wide flight from risk.

FAQ

What is the Philadelphia Semiconductor Index (SOX)?

The SOX tracks roughly 30 major U.S.-listed semiconductor design, manufacturing and equipment companies - including Nvidia, Broadcom, AMD, Micron and Intel - and is widely used as a barometer for the health of the broader U.S. chip industry, AI chips included.

Why does it matter that short sellers made $32 billion?

That figure represents the estimated unrealized paper gains held by investors with short positions (bets that a stock will fall) during the losing streak. A large number signals heavy capital positioned for the decline to continue - but it's a double-edged signal, because if the selloff stalls or reverses, those same short sellers often have to buy back shares to limit losses, which can fuel a sharp rebound.

Is this selloff a repeat of the dot-com crash?

By Bank of America's reading, chip stocks' valuation multiples have fallen into roughly the same range Cisco traded at after the 2000 crash - but there's not yet clear evidence that actual demand or earnings forecasts have turned down the way they did in 2000. That's a snapshot of the current moment, though, and the picture could change if real demand softens later.

Should I buy or sell chip stocks right now?

This article isn't investment advice. Whether valuations are compressed and whether earnings forecasts have actually deteriorated are two separate questions, and deciding how to act on them depends on each investor's own risk tolerance and time horizon.

Sources

This article is an original synthesis and analysis based on the reporting below, not a reproduction of the original articles. Please check the sources directly for the latest exact figures.

• Longest Chip Selloff Since 2006 Mints Short Sellers $32 Billion - Bloomberg
• Nasdaq sinks more than 2% as investors raise questions about AI debt - CNN
• Nvidia, Micron, Broadcom Trade Like Post-Crash Cisco: BofA - Benzinga

See also: CoreWeave Stock Crashes 16% as Bond Yields Hit 15%, CDS Signals 'Distress', September CPI Comes In Hot, Yet Semiconductor Stocks Fall for a Seventh Day

⚠️ This article is for informational purposes only and is not investment advice. Market conditions change constantly, so please verify the latest data before making any investment decisions.