Intel closed at $56 on 6 November, 60% below the $142 it reached on 22 June and back where it traded in March7. October was the worst month in the company's history, down 45%, and 30 October the worst day since 1974, down 28.6%56. The quarter that did it beat on revenue, margin and earnings. What the market sold was the guide: a fourth quarter of $15–16 billion against $17 billion expected, with the PC business down more than 20% sequentially and gross margin falling from 44% to 38%1. We think the market read the PC winter correctly and the rest of the company wrongly. The data centre grew 78% and is sold out through the first half of 2027, the 18A process reached industry-standard yield two years after everyone said it could not, and the balance sheet that nearly broke Intel in 2024 now holds $54 billion of cash with the US government, Nvidia and SoftBank as anchor shareholders. That is not a company in a crisis. It is a company that has just been repriced as one.
This piece belongs to our equities section, opened in November 2026 around one frame: the next cycle is being built in silicon and in qubits, and the companies that own the fabs, the tools and the machines are where the record will be written. The analysis is as of 9 November 2026.
Where we are now
The round trip is the story. Intel started 2026 near $37 after an 84% gain in 2025, fell 17% in January on a weak first-quarter guide, then doubled in April, up 114% in a single month, the best month in Nasdaq history, as 18A shipped in Panther Lake and Bloomberg reported Apple was in talks for foundry capacity87. It raised $23 billion of equity at $95 in August, upsized from $15 billion9. It peaked at $142 in June with a market value near $470 billion. Then the third quarter: revenue $16.5 billion, up 21%, non-GAAP gross margin 44.1%, up 200 basis points sequentially, earnings of $0.52 against a $0.38 guide1. And the fourth-quarter outlook that undid it all: revenue $15–16 billion, gross margin 38%, earnings $0.25 against a $0.43 consensus, client revenue down more than 20% quarter on quarter, a PC market management now expects to shrink 12% this year and 10% next, and a memory shortage it does not see easing before 202813. A fatal explosion at Fab 28 in Israel took another $300 million off the quarter1.
Inside the same release, the parts the market did not price. Data Center and AI revenue was $7.3 billion, up 78%, at a 42% operating margin, with server demand exceeding supply through the first half of 2027 and the industry's server CPU units growing 30% in both 2026 and 202713. Intel Foundry revenue was $6.7 billion, up 58%, and its loss narrowed to $1.85 billion from $3.2 billion a year earlier1. On the call the CFO said 18A yields had reached industry-standard levels ahead of internal projections and the CEO said they were improving at a double-digit rate; 18A wafer cost is down about 40% since the first quarter34. Clearwater Forest, the 288-core Xeon on 18A, is shipping for revenue. The 14A design kit went to select customers in October with the 1.0 release due in the first quarter of 20273.
At $56 the stock trades at 5 times sales, 2.6 times book and about 40 times depressed forward earnings; TSMC trades at 11 times sales7. The sell side cut targets by $20–45 in the week after the quarter and still averages $91, with a range of $63 to $160147.
The bull case
1. The data centre is sold out
Seventy-eight percent growth at a 42% margin is not a recovery, it is a different business. Granite Rapids is the largest server revenue driver and is now margin-accretive, Clearwater Forest is shipping, Diamond Rapids on 18A-P ramps in 2027, and management guided the segment up mid-teens again in the fourth quarter while saying it cannot fully meet demand13. Intel is losing unit share to AMD and Arm, 54.9% of server units from 64.4% a year earlier, and still growing the dollars 78%, because the market underneath is growing faster than the share is slipping13. Oracle signed a multi-year Xeon agreement in the quarter1.
2. 18A is working, and the bill has been paid
The thesis on Intel for three years was binary: either the process roadmap lands or the company becomes a design house. It landed. 18A is in high-volume manufacturing at Fab 52, yields are at the industry standard by the CFO's account, wafer cost is down 40%, and 18A-P entered risk production in June34. The first external leading-edge customers are the open question: Tesla on 14A was reported in October but not confirmed by Intel, Apple's reported agreement for an entry-level M-series on 18A-P from 2027 is unconfirmed by either side, and the only named external logic customer in the filings is Fortinet172. We treat those as unverified and say so. But the capital to get here is in the past: $23 billion of equity at $95, $5 billion from Nvidia at $23, $2 billion from SoftBank, $8.9 billion from the US Treasury at $2091110. The shareholders who funded the fab paid more for it than you can today.
3. The balance sheet is de-risked
Cash and investments of $54 billion against $50 billion of debt, a $15 billion undrawn revolver, net cash of $4.2 billion, headcount down from 101,000 to 81,000 in fifteen months, 2026 restructuring finishing this quarter12. Altera has filed for an IPO that could raise more than $2 billion with Intel still holding 49%7. The company that was being discussed as a break-up candidate in 2024 is now valued at 2.6 times book with the largest government, the largest AI company and the largest Japanese technology investor on its register, none of whom can sell quickly1011.
The bear case
Now the argument against. It is the one the market is pricing, and it has numbers.
The guide is the guide. Fourth-quarter revenue $15.5 billion at the midpoint against $17.1 billion expected, gross margin down 600 basis points, earnings cut by 40%, client revenue down a fifth in a quarter1. The PC market is shrinking 12% this year and 10% next by Intel's own estimate, the memory shortage that is strangling PC builds does not ease until 2028, and AMD's desktop share is at 35%312. Half of Intel's revenue sits in that market. Morningstar titled its note "PC winter has arrived".
Cash burn comes back in 2027. The CFO said it plainly: probably not positive free cash flow next year3. Capex goes from $20 billion toward $26–30 billion on analyst estimates, with equipment orders up 40% in a quarter4. Net cash is $4.2 billion. The share count is up 15% in a year, 241 million warrants and 132 million escrowed government shares are still outstanding, and the trailing net loss is $12 billion27. Dilution is how the fab got funded, and it is not finished.
No signed leading-edge volume, and a shrinking share. External foundry revenue is $411 million a quarter, most of it deconsolidated Altera4. Server unit share fell ten points in a year; Arm is heading for 20–30% of server units; Diamond Rapids slipped a year; Ohio slipped to 20301318. The quantum story is immaterial: Tunnel Falls is a 12-qubit research chip, the $2.3 billion Classiq acquisition closes next year and its own CTO says it will not be a high-margin business from day one1516. If a 14A customer does not sign by mid-2027, Intel has said it may pause leading-edge development2. That is the bear case in one sentence.
Intel has been repriced for the PC winter as if the data centre, the foundry and the balance sheet did not exist. The winter is real and we do not argue with it. But a company growing its largest profitable segment 78% with demand above supply, running a leading-edge process at industry yield, and holding $54 billion of cash funded by shareholders who paid $95, $23 and $20 a share, is not worth 60% less than it was in June. The fourth quarter is the trough the guide describes; 2027 is Diamond Rapids, Nova Lake, 14A customers and the Altera IPO.
Our view: Intel trades at $85, in an $80–90 band, by the end of Q4 2027. That is roughly the sell-side average and a third of the way back to June. Conviction is moderate-to-low — about 55% — and we say why: the stock has a beta of 2.4, the first quarter of 2027 carries the Fab 28 drag and the PC trough, and the foundry customer that would change the multiple is still a report, not a release.
Our bet runs one way: up, from a repriced base, on the data centre and the fab. We are not betting on Apple or Tesla being announced inside the horizon. If Intel closes a week below $45 first, we were wrong — see below.
What would prove us wrong
One direction, one closing level, and the fundamentals that would hollow it out.
$45 is where Intel closed on 23 January 2026 after its last bad guide, before the April rally. A week closing beneath it says the whole 2026 re-rating has been unwound and the market is pricing the fab as a liability again, and the thesis is closed against us, whatever happens afterwards.
Any of these weakens the thesis while the price holds, and we will log it as it happens: fourth-quarter results in January below the $15.0 billion floor of the guide; data-centre revenue growth falling under 30% for a quarter; 18A yield commentary reversing; or Intel announcing a pause in 14A development, which would mean the external foundry has failed to find its customer.
Intel trading in the $80–90 band on the NASDAQ by 31 December 2027. If neither the target nor the invalidation resolves by then, the thesis is booked as it stands — unresolved and closed. It does not get extended.
When a condition triggers, we book the result and publish the piece. As always.
This is analysis, not investment advice. Do your own research. The thesis is not changed after the fact.