Intel: PC Market Falters First, Foundry Plans Still "To Be Discussed Later"

On October 30th, Intel shares plunged nearly 30% during trading hours, just a day after the company reported its strongest data center growth in fifteen years. However, the market's drop was not about the current quarter's results, but about guidance for next quarter’s PC segment, declining margins, and the still-unmodeled external foundry business.

A few hours earlier on the earnings call, Bank of America skipped over the results and cut straight to pressing about the external foundry: When will clients move from evaluation to tape-out—finalizing design, sending it to the fab—and when will external revenues be able to be included in forecasting models? CEO Lip-Bu Tan said client details wouldn't be disclosed; on packaging, he stated “someone is preparing to announce in the next two quarters,” while regarding 18A and 14A he said “more information will come later,” without a timeline or modelable numbers.

Intel’s results for this quarter were not bad; the real problem was the upcoming quarter: revenue and EPS guidance both came in well below the generally expected consensus, with memory hitting the PC segment hard. Adding to the woes, there were more issues at Intel’s plant in Israel: Fab 28 in Kiryat Gat, one of the company's early 10nm production powerhouses still producing batches of 10nm server products, had a production halt for about two weeks following an explosion earlier this month.

Analysts pressed further: what will next year’s capital expenditures look like, and when will free cash flow turn positive? CFO Dave Zinsner said the figures aren’t locked in yet—next year is an investment year, and “probably won’t have positive free cash flow.” The stock tumbled nearly 30%, weighed down by slowing PCs, lower margins, and the “more information to come” stance on the foundry business.

Foundry Story Supports Valuation, PCs Drag Down Guidance

Intel once relied on its PC business. As it fell behind on process technology, former CEO Pat Gelsinger staked the company’s future on IDM 2.0: running its fabs not only for its own needs but also for external orders, piling money into manufacturing and chasing advanced nodes.

Lip-Bu Tan has been CEO for a year and a half; the market speculated he might cut back on foundry efforts to save the financials, but he didn’t—instead, he raised the priority for foundry even higher. Around August, the company issued another $23 billion in shares, padding its cash balance, and investors became more willing to bet on the “shortage easing, 18A ramping up, external clients paying” narrative.

This quarter's real problem was the Client Computing Group (CCPG). The client business (mainly CPUs for PCs and laptops) saw revenues fall to $8.097 billion, down 5% year-on-year and 9% quarter-on-quarter, missing company expectations, with operating profit of $2.285 billion and margins around 28%. It's still the largest segment by revenue, but volumes slumped first, and for Q4, the company guides for a further drop of over 20% sequentially. CFO also warned, “Looking ahead, we expect the PC slowdown to last at least through the first half of next year,” inventory drawdown will at least drag on until Q1.

The company's own industry forecast is even colder: PC TAM will drop about 12% by 2026, with H2’s decrease over 20%; 2027 will see another 10% drop. Memory price hikes are pushing up complete system bills and suppressing replacement demand. The same memory chip that data centers are still willing to pay extra for is now unaffordable for PCs, stalling shipments—Intel, as a CPU supplier, is hit first.

Morgan Stanley ran the numbers, ongoing deterioration in client business could drag consensus revenue forecasts down by $3-4 billion.

This Quarter Beats Yet Next Quarter a Concern

This quarter, Intel booked $16.513 billion in revenue, up 21% year-on-year and almost exactly matching consensus estimates near $16.5 billion; non-GAAP EPS was $0.52, well ahead of the $0.40 consensus and the guidance of $0.38.

Zinsner started the call: “We delivered another solid quarter... including the strongest data center revenue growth in fifteen years.”

Then came the problematic guidance: Q4 revenue guidance of $15-16 billion, versus consensus near $17.1 billion; non-GAAP EPS at $0.25, consensus about $0.43. Previously, some optimistic analysts pushed the Q4 revenue midpoint up to $17.1-17.5 billion, but guidance collapsed those forecasts immediately.

On the call, someone pressed: the guidance is about $1.8 billion below consensus, how much of that is from the Israel fab issue, how much from PC sales?

CFO Zinsner responded: according to preliminary estimates, the Israel fab incident accounts for about $300 million of Q4 revenue, almost all impacting 10nm server products, and drags down gross margin by about 200 basis points; Q1 revenue impact may be similar, with most wafer write-offs booked in Q4, but the biggest hit remains from weak PC sales.

Strongest Data Center Quarter, Still No Timeline for Foundry

This quarter, Intel's Data Center and AI segment (DCAI, mainly server CPUs and AI accelerators) posted revenue of $7.318 billion, up 78% year-on-year and 17% quarter-on-quarter; operating profit was $3.039 billion, with a margin around 42%, now exceeding CCPG in profit amounts.

Lip-Bu Tan said, “Despite near-term headwinds in the PC industry, CPU demand remains robust... Our top priority is unlocking extra supply to meet the extraordinary demand for the Xeon family.”

The supply bottleneck isn’t just at the fab level. Management noted Intel 3 output is climbing, but there’s a tight industry supply of advanced organic substrates, and EMIB advanced packaging is constricted—the wafers are ready, but packaging material lags, so server CPUs still aren't shipping fast enough.

The company expects server CPU shipments to increase by over 30% per year in 2026 and 2027, with bottlenecks possibly lasting through end-2027 or even into 2028; for the full year, management expects DCAI to beat initial forecasts by about $10 billion.

Against these robust server numbers, Morgan Stanley remains skeptical: the strong numbers rely heavily on shortages, and Intel’s flagship products remain relatively weak versus competitors.

The contrast with peers is sharper: on the call, it was highlighted that AMD’s server-related revenue guidance is about 70% higher, and Arm architecture is also taking server share. Rivals are talking revenue slope and market share, while Intel only says share will follow supply—however much supply improves, share will change accordingly. When there’s a lack of substrates and packaging, shortages create shipping pressure rather than automatically boosting market share. Demand is rising, but precise points remain unquantified.

This quarter’s foundry segment (mainly internal and external wafer manufacturing and packaging) booked revenue of $6.697 billion, almost all from internal orders; external foundry revenue was $411 million, up $118 million quarter-on-quarter, mostly from one-off client projects, with an operating loss of $1.853 billion.

TSMC remains the external foundry benchmark; only if clients keep paying for advanced nodes does the foundry story sustain.

SemiAnalysis noted that while 18A has shipped, its transistor density does not surpass TSMC’s comparable process, and high-end GPU and I/O chips still need to use TSMC capacity. Morgan Stanley was more direct: market faith in Intel’s foundry is mostly staked on 14A node by 2028 or 2029, but current high-volume external client commitments are still small, and the greater concern is that Intel’s structural capabilities are insufficient. If it cannot even meet its own demand, that would be the worst possible advertisement.

Cash Awaits Foundry, Foundry Timeline Still Missing

In August, Intel issued about $23 billion in new stock, ending the quarter with roughly $54 billion in cash and short-term investments; with diluted shares about 5.3 billion, Q4 guidance using fully diluted 5.44 billion shares, and interest-bearing debt totaling about $50.1 billion. Cash is abundant, but what’s missing is when external foundry revenue will become modelable.

Bank of America kept coming back to this question in Q&A. Lip-Bu Tan made clear in his remarks that client details won't be disclosed; when pressed about tape-out timing and modelable revenue, the answer remained, “more on this later.”

Capital expenditures are expected to remain around $20 billion in 2026, and “rise sharply” in 2027. The company hasn’t committed to any fixed numbers, and the circulating larger estimates are only rough approximations. Zinsner was blunt: next year is very unlikely to be a positive free cash flow year.

Single-quarter adjusted free cash flow was still $1.6 billion—operating cash flow was about $5.1 billion, minus gross capex of $4.2 billion, plus government incentives and partners’ contributions—this positive figure won’t hold up next year. The money came from operating receipts and the August share issuance, and is mainly being spent on wafer and packaging lines, as well as the still-lossy foundry division. With this financial model, Morgan Stanley concludes that even if numbers are revised up later, the stock is likely already fully valued at current pricing.

Server orders are still piling up, AI demand hasn’t tanked, but pricing is changing: PCs are stalled by rising memory costs, margins are being guided downward, external foundry revenue remains unmodeled, and next year’s free cash flow is likely negative again. The market no longer wants to pay for an unfulfilled foundry timeline and uncertain PC recovery.