INTC (Intel Corp) — Framework and Thesis #

Status: OPEN, 14 sh @ USD 90.10 avg (3 @ 89.00 on 2026-08-27, 11 @ 90.40 on 2026-08-31; EUR 1,093.39). Bought without a thesis; this document is the thesis written after the fact on 2026-09-02 at USD 89.81. Data: stock-data-analyst pass 2026-09-02 (SEC 8-K/10-Q, yfinance, Finnhub); fact-check pass same day (EDGAR Form 4 and 8-K, earnings transcripts, House Clerk PTR PDFs). Plan: INTC_plan.json. Execution log: INTC_execution.md.

1. How to judge this business #

Intel is a turnaround in a capital-intensive industry: it is rebuilding its manufacturing lead (18A, then 14A) while its products lose share, and it is paying for the fabs with other people's money. Graham's earnings formula does not work at a loss-to-profit inflection (rule-004), so the framework is four questions, each with numbers:

2. The three things that were true when it was bought #

Intel 18A is on track, by Intel's own statements (this is the node the owner called "18nm"; it is a 1.8nm-class process). Q2 2026 call, 23 Jul 2026: "18A output increased meaningfully in the quarter. Yields continue to track ahead of expectations." Panther Lake is in high-volume manufacturing for "a subset" of SKUs, with unit cost down ~50% year-to-date and 20% more planned. Clearwater Forest (Xeon 6+, first 18A server part) launched 2 Jun 2026 at the end of its revised window and is available from Dell, HPE, Lenovo and Supermicro. 18A-P entered risk production on time. 14A: PDK 0.9 due October, risk production 2H 2027, volume 2028. In July Intel deleted the 10-K language that it might cancel 14A without an external customer and said it has "committed to completing" it.

The CEO buys and does not sell. Lip-Bu Tan: 1.04M shares at USD 23.96 in March 2025 (USD 25M, required by his offer letter) and 105,263 shares at USD 95.00 on 11 Aug 2026 (USD 10M, purchased inside the public offering at the offering price, not on the open market). No sales. CFO David Zinsner bought USD 250K at 42.50 in January 2026 and has sold nothing. The only sales in twelve months: a departing Chief Legal Officer (USD 5M), the retiring chair (USD 9M, Form 144s), and the Foundry chief trimming ~9% of his holding (USD 2.5M). No cluster. The owner's insider test passes today.

The Pelosi trades are real. House Clerk periodic transaction reports: 29 May 2026, spouse bought 200 INTC call options, USD 50 strike, expiring 19 Mar 2027, USD 1–5M; 24 Jul 2026, 50 more USD 50 calls (USD 250–500K) and 10,000 shares (USD 500K–1M). No sales disclosed. Two things to understand about this signal: deep-in-the-money calls bought with the stock at ~USD 115–120 are a leveraged long, not information about 18A yields; and the filing reached the public 25–45 days after the trade, after the stock had already moved. Rule-001: this is attention, not analysis. The CEO's Form 4 is the insider signal; a member of Congress is a follower like everyone else.

3. Good? (rule-003) #

One pass, two ambers, two fails. Intel is a turnaround that has cleared its first milestones, not yet a good business by this checklist: it does not fund itself and it dilutes.

4. Cheap? (rule-004) #

Price USD 89.81; 5,285M shares; market cap ~USD 475B; EV/sales ~8.3x; P/B 4.3x, P/TBV 5.4x (post-offering) on ~0% return on equity.

Peers: AMD 41x forward, TSMC 19x (55% gross margin, 30%+ ROE), GlobalFoundries 20x. Analysts: 4 strong buy, 15 buy, 34 hold, 2 sell; targets 75 / 115 mean / 200 high; JPMorgan Underweight at 85.

What USD 90 needs under V = EPS × (8.5 + 2g): at 10% sustained growth, EPS 3.15 (net income ~USD 17B); at 15%, EPS 2.33 (~USD 12B). Consensus FY2027 is 2.04. So the price assumes FY2028 earnings of roughly USD 12–17B on a company that earned about zero operating income in the last twelve months. That is the bull case for 18A and a 14A customer, already in the price.

Scenarios on FY2028 non-GAAP EPS (assumptions, stated so they can be argued with):

Verdict: not cheap, and not fairly priced either. The expected value is below the price; the margin of safety is negative, which is the same verdict the June review reached before the position was bought. Kelly as a go/no-go (rule-006): no edge, no adds.

5. Sizing (rule-006) #

EUR 1,093 is 34% of the single-name book, the largest position, above the 20% cap. The cap is a rule independent of the thesis: the position should be trimmed to 8–9 shares (≈ EUR 620–700, 20–22%). The rung ladder below then works on the remainder. This is the owner's decision; the plan records it as a pending action, not an auto-sell.

6. Kill conditions (fundamental only, rule-008) #

1. Insiders: any open-market sale by the CEO or the CFO, or two or more C-suite discretionary sales within 90 days. Owner's rule.
2. 18A: Intel states a slip of a quarter or more in Panther Lake volume, reports 18A yields below plan, or withdraws the "gross margin comfortably in the 40s" language.
3. 14A: no external customer commitment by mid-2027 (Tan's own window: second half of 2026 to first half of 2027), or 14A paused.
4. Dilution: any new equity issuance above 2% of shares, or the government warrant triggered (foundry stake below 51%).
5. Non-GAAP gross margin below 38% for two consecutive quarters.
6. Foundry operating loss not narrowing year-on-year for two consecutive quarters, or external foundry revenue flat.

Any one fires: re-score within five trading days; exit unless the thesis survives on the new numbers.

7. Take-profit ladder (rule-008, mechanical: no margin of safety cushion) #

8. Monitoring #

• Next print 2026-10-21 after close (consensus revenue USD 16.7B, non-GAAP EPS 0.39): Panther Lake volume language, gross margin vs the 42% guide, foundry loss, capex, any 14A customer news, share count.
• Form 4: CEO and CFO purchases are the signal; any sale by either is kill condition 1.
• Government: escrow releases, warrant status, any change to the CHIPS dividend/buyback restrictions.
• Short interest is 2.6% of float with 1.2 days to cover: no squeeze mechanics, nothing to read there.
• Congressional disclosures are lagging attention (rule-001); do not act on them.