Intel, Marvell and Micron each fell sharply this week while Nvidia’s own stock initially held up better, a pattern more consistent with investors trimming the semiconductor sector’s most extended gains than with a broad repricing of the AI-spending story ahead of Wednesday’s report.
Semiconductor stocks sold off broadly on Monday, two sessions before Nvidia’s Wednesday earnings report, and the selling was not evenly spread. Intel fell 5% to $85.98. Advanced Micro Devices fell 4% to $454.36. Marvell Technology fell 4% to $228.31. Micron fell 7%, even though the company has already sold out its 2026 memory output. Taiwan Semiconductor fell 3% to $406.40. Broadcom fell 2% to $360.49. The iShares Semiconductor ETF fell as much as 4% to $501.17, roughly four times the 0.7% decline in the Invesco QQQ Trust the same day, and more than the Nasdaq Composite’s 0.44% decline. The Dow Jones Industrial Average rose 0.27% the same session, evidence of rotation into blue chips rather than a broad flight from equities.
Nvidia itself moved less than most of its peers on Monday, down 1.9%, before extending its decline to 2.91% on Tuesday, the day before its report, as the print drew closer. The stock’s market capitalization stood at $5.05 trillion Tuesday, still the largest in the index.
Two features of the selling argue against reading it as a verdict on artificial-intelligence spending itself. First, it landed hardest on the sector’s most extended year-to-date gainers: Intel was up 144% for the year through Friday, and Marvell was up 179%, making both natural sources of funds when investors reduce exposure ahead of a binary catalyst. Second, Micron’s seven percent decline came despite the company having already sold its entire 2026 memory output, a fact that argues against the drop reflecting a change in near-term demand for that specific company. A sector-wide move that hits a company with no near-term demand question is more consistent with positioning and profit-taking than with a reassessment of end demand.
Nvidia reports Wednesday against a consensus of roughly $91.9 billion in revenue and $2.08 in earnings per share, with one widely cited estimate set slightly higher at $92 billion and $2.10. Wall Street’s own next-quarter consensus already expects acceleration, to roughly $104 billion in revenue and $2.37 in earnings per share, and analysts are watching gross margin against a roughly 73.5% benchmark for signs of pricing pressure from custom AI chips built by cloud providers themselves. China exposure is a live variable: export approval for Nvidia’s H20 chip has come through, but no orders had been confirmed as of this week. Nvidia has beaten consensus in each of its last four quarters, and its shares fell on the day of the report each time regardless, a pattern that has itself become part of what investors are pricing.
That history changes what the report needs to do. A number in line with consensus, delivered into a stock that has already given back some of its gains, is not obviously enough on its own to reverse the week’s positioning; the market’s reaction is more likely to hinge on the size and tone of forward guidance than on the headline beat. A guidance number below the roughly $104 billion already expected for the following quarter would extend the pattern that has followed Nvidia’s last four reports. A number meaningfully above it, particularly paired with confirmed China orders, would be the clearer signal that suppliers, not Nvidia, absorbed this week’s de-risking.
None of this establishes that Nvidia’s earnings caused this week’s selling; the selling started before the report existed to react to. What the pattern does show is that the semiconductor complex has been trimming its most extended winners into a known catalyst, while the company most directly exposed to that catalyst’s outcome fell the least until the print was almost on top of it. Whatever Nvidia reports Wednesday, this week’s declines in Intel, Marvell and Micron will not be explained by Nvidia’s own numbers alone.