Stock X-Ray: Intel (INTC) — full dossier

Research and education — not personalised buy/sell advice. Figures as of 28 October 2026. Prices: Nasdaq official daily history. Financials and guidance: SEC EDGAR filings and Intel earnings releases (Ex-99.1). Aggregator multiples labelled SECONDARY.

I. Verdict

Intel's six years split cleanly by chief executive. The Swan window quarter was a cash-return business beating its own guidance. The Gelsinger era spent $87.4 billion of capex chasing process leadership, burned $25.4 billion of free cash flow, and lost 61% of the share price while the market rose 53%. The interim quarter stabilised the company. The Tan era has so far been a guidance-and-repricing story: revenue beaten against guidance in six of six quarters and the share price up 265% off a crushed base — with free cash flow still negative and GAAP EPS negative in five of Tan's six quarters.

Valuation label: OVERVALUED (vs the industry, on earnings) — explicitly not BUBBLE. Forward P/E ~54× headline (~61× on FY2026E non-GAAP EPS of $1.54) vs a peer median of 19.7× and Intel's own 14–29× forward range across FY2021–24. The price requires execution to keep compounding, not demand to accelerate — demand is real and primary-sourced — which is overvaluation risk, not bubble structure. Full evidence in §III.

II. The IT factor — foundry potential, priced at zero

The IT factor: foundry potential — the most exciting thing about Intel, and the part the market is pricing at zero.

External foundry revenue was roughly $0.3B of Intel's $16.1B in Q2 2026 (Intel Foundry segment revenue $5.8B less $5.5B of intersegment eliminations — i.e., under 2%, primary: Q2 2026 release), so consensus models — and the current analyst fight, which clusters around Hold targets of $98–$115 on CPU share, margins and near-term EPS (SECONDARY) — carry the foundry option at effectively nothing. It is overlooked because foundry reaches the P&L today only as losses and internal transfers, and because six years of process delays trained the market to pay for Intel foundry only when a customer is named — and Lip-Bu Tan has never named one on an earnings call (six calls studied). Meanwhile the asset is real and de-risked in a way no normal turnaround is: 18A is in volume production, the 14A PDK path is rolling (0.5 out, 1.0 due early next year — as reported from Tan's Oct 24 CASPA remarks), and the US government sits on the shareholder register. The materiality test is binary: one named hyperscaler-scale external commitment converts a zero in every model into a multi-billion-dollar, incumbent-multiple revenue stream flowing into fabs whose fixed costs Intel already carries — that single headline re-rates the stock more than any CPU quarter can. If the 14A window passes with still no named customer, the option decays and Intel is just an expensive CPU recovery.

III. Valuation

Prices as of the Oct 27, 2026 close. Multiples: stockanalysis.com (SECONDARY, S&P Global data), Oct 28, 2026.

Peer forward P/Es (Oct 27 closes, SECONDARY):

Read of the table: INTC's forward earnings multiple is ~2.7–3.1× the peer median and above every peer except AMD (the other name priced on an AI-CPU recovery). On price/sales it is the second-cheapest of the six — the expensiveness lives entirely in the missing "E", not in the revenue line.

Sector demand read (x86 CPU + leading-edge foundry). Demand for compute is rising and primary-sourced: Intel's Q2 2026 release reports revenue +25% YoY with Data Center & AI +59%, states "AI is driving unprecedented demand for compute," and discloses Intel is increasing investment in equipment, clean-room space and substrates "to support expected growth this year and next across products and foundry." Hyperscaler CPU demand got a fresh datapoint on Oct 27, 2026: Intel and Oracle announced a planned multiyear expansion of Xeon on Oracle Cloud Infrastructure (Business Wire, Oct 27) — products demand, not foundry. In the Bottleneck Intelligence module's standing view (SIP-2026-001), the AI cycle's scarce layers are substrates, packaging (TSMC CoWoS), EUV, power and cooling; leading-edge wafer capacity is concentrated at TSMC, and neither x86 CPUs nor Intel Foundry is a ranked bottleneck candidate — Intel is a would-be second source into a concentrated market, not the bottleneck owner. Net read: sector demand rising; foundry demand real but currently routed to the incumbent.

Why OVERVALUED, and why not BUBBLE. (1) ~54–61× forward earnings vs a 19.7× peer median and Intel's own 14–29× FY2021–24 range; the stock re-rated +145% in 52 weeks (SECONDARY) while GAAP TTM EPS is still negative. (2) The multiple is carried by an assumed earnings normalisation: FY2026E non-GAAP EPS is $1.54; the price needs EPS to roughly triple within ~two years to compress toward peer multiples — a forecast, not a run-rate (Q3 2026 non-GAAP guide is $0.38). (3) Dilution is part of the price of survival: shares +13.28% YoY. Not BUBBLE because the demand underneath is real and primary-sourced, P/S at 8.6× is mid-pack-to-cheap rather than extreme, and a genuine un-modelled option (foundry) accounts for part of the premium — the price requires execution to continue, not demand to accelerate. Not FAIRLY PRICED because at ~3× the peer-median forward multiple with the weakest margins in the set (TTM gross margin 38.9% vs peers' 56–81%, SECONDARY), "fair" would require the recovery and the option both to be near-certainties.

Falsifier (what changes the label): Up to FAIRLY PRICED if either (a) a named external foundry customer commits at scale on 18A/14A, or (b) delivered non-GAAP EPS annualises toward ~$3 (the Oct 29 Q3 print + Q4 guide is the first hard test). Down toward BUBBLE-risk if the Q3/Q4 prints show the revenue surge stalling, EPS sticks near ~$1.5, and no customer is named by the 14A 1.0 PDK window (early 2027) while the price holds.

Is the P/E rational? At ~54–61× forward earnings, nobody is paying for Intel's current earnings power — on delivered numbers the multiple is indefensible, full stop. What the price actually buys is a normalised Intel — CPU share defended inside a genuinely rising AI-compute market, gross margin climbing back from 40% toward historical levels, plus a foundry option the models carry at roughly zero — and that is a coherent price, not an irrational one: the demand evidence is primary-sourced, the guidance record under Tan is six beats in six quarters, and sovereign backing makes the multi-year runway financeable. So the P/E is rational as the price of a recovery-plus-option, and expensive as the price of a business. Its rationality is entirely conditional: it survives only if EPS roughly triples from the FY2026E base within about two years and the foundry option stays alive; if either leg fails, there is no earnings floor under $93.74 anywhere near the current price.

IV. The eras

Era rule: a fiscal quarter is assigned to the CEO in office on the quarter's final day. Boundaries are verified from Intel press releases furnished as Exhibit 99.1 to Form 8-K: Gelsinger appointment effective Feb 15, 2021 (accession 0001193125-21-009142); Gelsinger retired effective Dec 1, 2024, Zinsner & Holthaus interim co-CEOs (0000050863-24-000173); Tan appointed effective Mar 18, 2025 (0000050863-25-000036). Swan became CEO Jan 31, 2019 — only his final quarter falls inside the six-year window.

Price returns are on Nasdaq official daily closes, INTC vs SPY indexed to 100 at Sep 30, 2020 (INTC $51.78, SPY $334.89); dividends not reinvested. Full window: INTC +81.04% vs SPY +129.82%.

Era colour: Gelsinger also paid $16.33B of dividends and $2.42B of buybacks across his era while FCF was −$25.42B; Swan's single window quarter returned $2.00B of buybacks and $1.35B of dividends. The Tan-era +264.89% starts from a crushed base ($25.69) — the level, not just the percentage, matters. Tan-era GAAP EPS by quarter: −0.19, −0.67, +0.90, −0.12, −0.73, −2.16 — negative in five of six.

V. Earnings record — guidance vs actual

Scored on revenue guidance vs actuals only (historical analyst consensus 2020–2024 is unavailable from primary sources without a paid terminal — stated limitation). Verdict rule: BEAT = actual more than $0.1B above the guidance midpoint (or point guide); MET = within $0.1B; MISS = more than $0.1B below. A miss vs midpoint can still be inside the guided range. Tallies: Swan 1 BEAT / 1 · Gelsinger 8 BEAT / 3 MET / 4 MISS of 15 · Interim 1 BEAT / 1 · Tan 6 BEAT / 6. Full 23-quarter table in the Appendix.

VI. Judgement

• Swan: a cash-return business beating its own guidance; the window quarter is too short to grade a strategy.
• Gelsinger: $87.4B of capex bought process ambition and a −72% drawdown; revenue fell every year of the era and GAAP gross margin ended at 15.03% in the Q3 2024 impairment quarter.
• Interim: stabilisation, not strategy — one quarter, guidance beaten, price +6.8% against a falling market.
• Tan: six-for-six guidance beats and a stated under-promise/over-deliver pattern; but FCF is still negative (−$3.04B cumulative) and GAAP EPS was negative in five of six quarters — the re-rating is running ahead of the delivered P&L.

VII. Outlook

Q3 2026 reports Thu Oct 29, 2026 after close (call 2pm PT / 5pm ET) — Tan's seventh quarter on this scorecard. The verified bar is Intel's own Q3 guide from the Q2 2026 release (accession 0000050863-26-000155): revenue $15.8–16.8B, non-GAAP EPS $0.38, GAAP EPS $0.31, non-GAAP GM 42.0%. Street consensus (~$16.5B / ~$0.40, from Oct 22–23 research) is NOT re-verified as of Oct 28 — flagged UNVERIFIED.

VIII. Limitations

• Q3 2026 consensus figures carried from Oct 22–23 research are UNVERIFIED at this as-of date.
• Historical analyst consensus (2020–2024) unavailable from primary sources without a paid terminal; the earnings record scores guidance vs actuals only.
• Buybacks FY2024–26 rest on a documented zero-fill basis (no repurchase line in EDGAR company facts); dividends 2025–26 are $0 under the Aug 2024 suspension.
• FY2027 consensus EPS UNVERIFIED (see §III). 14A PDK milestones rest on a single reported CASPA clip transcript (Oct 24, 2026), not an Intel release.
• The ~$0.3B external-foundry figure is arithmetic on the Q2 2026 release's segment table; Intel does not print a clean external-foundry revenue line.

IX. Sources

SEC EDGAR: Intel 8-K Ex-99.1 releases (accessions 0001193125-21-009142; 0000050863-24-000173; 0000050863-25-000036; 0000050863-26-000155), EDGAR company facts (financials), DEF 14A (accession 0000050863-26-000061). Nasdaq official daily price history (INTC, SPY). stockanalysis.com statistics/forecast pages (SECONDARY, S&P Global) for multiples and peer set, fetched Oct 28, 2026. Business Wire, Oct 27, 2026 (Intel–Oracle). SIP-2026-001 (Bottleneck Intelligence founding study) for the sector demand frame.

Appendix — full quarterly detail (23 quarters)

Financial history (revenue, gross margin, GAAP EPS, free cash flow, capex) and the full guidance-vs-actual record follow in the tables below.

Financial history — 23 quarters