Intel's 2026 rally hasn't collapsed because investors suddenly stopped believing in AI. It has hit trouble because AMD is taking real data-center dollars, Intel's 18A factory story still has margin questions, and the chip trade has become too hot to treat every dip as noise.
Intel closed Friday, July 10, at $109.84, according to MarketWatch, 22.84% below the 52-week high of $142.35 it reached on June 30. That's a hard turn for a stock that had spent much of 2026 looking untouchable. Then it stopped. You don't lose that much ground in less than two weeks because one analyst changed a spreadsheet. You lose it when investors start asking whether the turnaround story is still moving faster than the facts.
AMD is no longer waiting outside #
The cleanest warning came from AMD. MarketWatch reported that AMD's first-quarter 2026 data-center revenue rose 57.2% from a year earlier to a record $5.78 billion, above the FactSet consensus estimate of $5.64 billion. Intel's own Data Center and AI unit brought in $5.1 billion in its latest quarter, up 22%, also according to MarketWatch. Both companies are growing. Only one is changing the balance of power.
That's the uncomfortable part for Intel. For years, its server business was the thing investors could fall back on while everything else looked messy. AMD was the challenger, the discount alternative, the company you watched because Lisa Su had made it impossible to ignore. Now AMD is taking the kind of data-center revenue Intel used to treat as its home turf. First time. If you're buying Intel as a comeback story, you can't wave that away.
Server market share tells the same story in a rougher way. Recent reports have put Intel's server CPU share in the high-60% range, down from the low-70% range a year earlier, with AMD and Arm-based designs picking at the edges. Intel still leads. That's real. But the lead is thinner, and in a market priced for perfect AI demand, thinner is enough to hurt.
The factory bet still has to prove itself #
Intel's foundry story is where the stock gets its upside and its risk. Tom's Hardware reported that Intel's 18A process is progressing, but management has said yields are still not where they need to be for normal margins, with desired cost levels expected by the end of 2026 and industry-standard levels in 2027. That's not a fatal problem. It is a timing problem, and timing is exactly what a stock at record highs doesn't forgive.
Intel has pushed back with its own progress story. Chief executive Lip-Bu Tan has said 18A is on track, and finance chief David Zinsner has pointed to improving yield volatility. The company has also talked up 14A, the node after 18A, as a bigger foundry opportunity. Fine. But investors don't get paid on roadmaps alone. They get paid when fabs produce chips at margins that make the capital spending make sense.
There was also fresh news on Monday, July 13. The Wall Street Journal reported that Intel plans to invest 5 billion euros, about $5.71 billion, to expand manufacturing at its Leixlip facility in Ireland, with the money aimed at production and research capacity tied to AI and high-performance chips. That is the right kind of industrial commitment if you believe Intel can rebuild its manufacturing edge. The awkward detail is that Intel shares still fell in early trading. Investors noticed the spend before they rewarded the story.
The AI trade is no longer getting a free pass #
The broader chip market gave Intel no cover. Investopedia reported on July 8 that chip stocks were set to extend losses after Samsung's preliminary second-quarter results beat estimates but failed to impress investors, while renewed AI bubble worries weighed on Nvidia, Intel, AMD, Micron and other hardware names. The iShares Semiconductor ETF was down in premarket trading that morning. This is what happens when expectations get too clean. Good numbers stop being enough.
MarketWatch put the move in sharper terms, noting that the PHLX Semiconductor Index fell 4.7% on July 7 and that SOXX had declined 16% from its late-June peak. Intel was hit harder than most, but it wasn't alone. AMD, Micron, Samsung, Taiwan Semiconductor, you name it, the market started treating AI hardware as a crowded trade rather than a one-way machine.
Frankly, that's the split you have to watch. Intel's bulls can still make a serious case: data-center demand is strong, server CPUs are becoming more important as AI shifts toward inference, and a working 18A ramp would change how investors value the company. The bears don't need to say AI is fake. They only need to say Intel's factories, margins and competitive position aren't ready for the valuation investors gave it in June.
The clock is short. Intel's next earnings report is the test. It will show whether 18A is moving toward profitable production, and whether Data Center and AI can keep growing while AMD presses harder. If the company gives investors real yield progress and strong server demand, the recent drop can look like a harsh reset. If it doesn't, the story changes. The stock's record rally will look less like a comeback and more like a market that got ahead of the work still left to do.
Also read: Klarna Wants Its Own Bank, and Regulators Now Have to Decide • Wall Street's biggest banks are lobbying to gut the law that legalized stablecoins • Saudi Aramco Bets $800 Million That Open Source Beats Closed AI