2026-09-16 — SK Hynix explores leasing Intel’s Ohio fab, or a joint venture, for US memory production

• SK Hynix is in exploratory talks with Intel about a deal to manufacture memory chips on US soil for the first time, Reuters reported 09-16 (exclusive), under one of two possible structures: SK Hynix leasing part of Intel’s Ohio One campus, or a joint venture between Intel, SK Hynix, and major cloud providers seeking guaranteed memory supply. SK Hynix told Reuters it is “reviewing options” but that “no matters have been determined”; Intel declined to comment on the talks beyond saying it continues investing to prepare the Ohio site. There is no MOU, capacity commitment, or process-node disclosure from either side, and the timeline is long: Intel’s first Ohio module isn’t scheduled for construction completion until 2030, with operations expected 2030-31. This is new information on this thread’s “Ohio’s fate” watch, distinct from the 07-21/22 SK Hynix buyout rumor that SK Hynix denied — a narrower lease/JV structure for memory (not logic) production, not the campus purchase reported (and denied) two months ago. (Reuters, original, Windows Forum, detailed writeup)
• A GF Securities analyst report (Jeff Pu), covered 09-15/16, estimates Intel’s 18A process yield has climbed to approximately 80%, up from about 65% in Q2 2026 — the first specific yield percentage on record for this thread, which has flagged the absence of any yield number since it opened and could only log the CFO’s qualitative “best since 22nm” description on 08-28. The same report projects Intel’s next-generation 14A process reaching about 6,000 wafers/month capacity in 2027, scaling to 24,000/month by 2028 (excluding Tesla/SpaceX’s separate Terafab line), and raised Intel’s price target to $150 citing the 18A/EMIB-packaging yield improvement and continued foundry-customer growth. This is an analyst estimate, not a company disclosure — flagged as such — but it is the most concrete yield data point available and directly fills the gap this thread’s own watch line named. (BigGo Finance)

2026-09-02 — Intel reverses course on selling its Folsom campus, choosing to keep the asset rather than raise capital against it

• Intel confirmed Wednesday it has scrapped its November 2024 plan to sell-and-partially-lease-back its 161.8-acre, ~1.67 million-square-foot Folsom, California campus (data centers, labs, chip-verification equipment, occupied since 1984), saying in a direct statement: “After careful consideration, Intel has decided not to move forward with the sale and partial leaseback of our Folsom site. We remain committed to the Folsom campus and have determined that retaining ownership best supports our long-term objectives.” The original plan, first reported by the Sacramento Business Journal, would have sold the property while leasing back roughly 1.02 million sq ft under a 12-year term — a capital-raising move typical of Intel’s 2024–2025 cash-preservation era (workforce fell from ~100,000 to under 78,000 over the same period). Reversing it now, under CEO Lip-Bu Tan’s shift toward core computing/manufacturing/foundry emphasis and against a backdrop of AI-driven chip demand lifting Intel’s stock, is a real if modest data point on this thread’s “subsidized-builder paradox” watch: a company that took $10.9B of outside equity to stay solvent eighteen months ago is now declining to monetize a real-estate asset it once needed to sell. (Folsom Times, primary quote, The Real Deal, Fox40)

2026-08-28 — Intel’s 14A process posts its first real yield data point — defect density “best since the 22nm process,” per the CFO

• Intel’s CFO said 14A defect-density improvement is the best the company has seen since its 22nm node, and that customers have shifted from “evaluating data” to “competing for capacity” — filling a gap the thread has flagged since it opened (“no yield % anywhere”). Corroborated across Tech Times, finance.biggo.com, and igor’sLAB (which ran the actual defect-density curve against the 22nm comparison). Stock reaction was muted-to-negative on the same day (-2.9%) alongside a reported $2.1B foundry-unit loss, so this is a real signal inside continued financial pressure, not a turnaround headline. (finance.biggo.com, Tech Times)

2026-08-11 — The offering upsizes to $20B on 5x institutional demand

• Intel upsized the offering logged yesterday from $15B to $20 billion, pricing 210.5 million shares at $95 each after institutional demand topped $100 billion — roughly 5x oversubscribed — with underwriters holding a 30-day option on another 31.6 million shares. Net proceeds land near $19.7B, still earmarked for “general corporate purposes” including foundry capex; the offering closed 08-12. (Intel Newsroom, CNBC)

2026-08-10 — Intel raises $15B in fresh equity for the foundry buildout

• Intel launched a $15 billion common-stock offering to fund its foundry buildout, J.P. Morgan, Goldman Sachs, Morgan Stanley, and Citigroup as joint bookrunners, with a 30-day underwriter option for another $2.25B. A private capital raise stacked on top of the CHIPS-Act equity stake and the Nvidia/SoftBank investments this thread already tracks. (Intel Newsroom)

2026-07-26 — The stake, marked to market: $8.9B → $36-42B

• The government’s Intel stake is now worth $36-42B, up from the $8.9B paid — cited alongside a wider, uncoordinated federal equity portfolio: ~$26.7B across ~30 equity/quasi-equity deals scattered across Commerce (17), Defense (7), DFC (6), and Energy (2), with no centralized oversight (no TARP-style inspector general or GAO audit equivalent). Directly answers this thread’s “does the state stake change behavior” watch: so far, no — it’s accumulating with no visible governance layer. (Fortune)
• ⚠ Contradiction, unresolved — three figures, one stake: Trump says the position “made about $70 billion in eight or nine months”; a separate Yahoo Finance/Treasury-sourced figure puts unrealized gains at “roughly $47.6 billion” (implying ~$56.5B current value); Fortune/TheNextWeb/Biggo separately size the whole position at $36-42B. None of the three reconcile with each other on the same $8.9B cost basis. Unresolved: the real number is not established. (Yahoo Finance)

2026-07-23 — Q2 2026: the best quarter in ~15 years

• Revenue $16.13B (+25% YoY, fastest growth in ~15 years), adjusted EPS $0.42 vs $0.21 expected — AI-driven business (Foundry + AI-adjacent products) grew 70%+ YoY and now runs ~70% of revenue; gross margin 41.8%, ~280bp above guidance. Stock fell 2.46% in the regular session to $100.10, then rose 3.44% after-hours as investors digested the beat. 2026 capex guidance raised to $20B+. (TradingKey)
• This is the first real revenue evidence behind the turnaround thesis this thread tracks (18A/14A execution, foundry customer wins) — thrust was reading ≈0 in the crawl; today’s print is the first print that argues otherwise, pending confirmation the AI-segment growth holds.

← Backstory

Backstory — the stake-value tangle, resolved 2026-07-28

Finding: artifacts/findings/intel-stake-tangle-2026-07-28.md · Bundle: artifacts/bundles/intel-stake-tangle-2026-07-28/provenance.yaml

• The three public figures are one position measured three ways — fixed basis 433.3M sh @ $20.469 = $8.87B: $36B = April price · $42B = July price (Fortune, the best current figure) · ~$56.5B = the May-8 price peak ($129 intraday, Apple-deal day; a Kobeissi Letter number). Trump’s “$70B” gain is unsupported — needs ~$182/sh stock-only (never traded), reachable only by counting the warrant’s 240.5M shares at full spot with the $20 strike and non-exercisability ignored, at pre-selloff prices.
• The filing-accurate number nobody uses: the government holds 290.3M shares today, not 433.3M — 143M still escrowed against future Secure Enclave disbursements (Q2 FY26 10-Q). Literal current value $25-28B, gain ≈ $19-21B. Ownership diluted to 8.6% (5.8% issued-only) by the SoftBank (Aug-25) and Nvidia (closed Dec-25) raises — the “9.9%” is a 2025 headline.
• No SC 13D/13G was ever filed for the position — 8-Ks and press releases only; unusual for a >5% holder, and the disclosure-side twin of Fortune’s “in no budget document” finding.

Backstory — reconstructed 2026-07-28

Finding: artifacts/findings/intel-rescue-2026-07-28.md · Bundle: artifacts/bundles/intel-rescue-2026-07-28/provenance.yaml

• SoftBank stakes Intel first, $2B — announced 2025-08-18/19 (SPA executed 08-28, closed 09-26), 86,956,522 shares at $23.00/share, widely read as Son personally backing CEO Lip-Bu Tan. (SEC 8-K)
• The layoffs and the fab cuts land together — 2025-07-24/25: ~24,000 layoffs announced (target 75,000 “core employees”; Tan: “no more blank checks”), Magdeburg Germany fab cancelled outright, Poland assembly/test site scrapped, and New Albany, Ohio construction “further slowed” — all in the same Q2 2025 earnings-adjacent announcement wave. (Google News RSS)
• US government takes a 9.9% stake, $8.9B, with a foundry-contingent warrant — agreement 2025-08-22, closed 08-26: $5.695B in accelerated CHIPS Act funds + $3.175B Secure Enclave funds converted to 433.3M shares. A separate 240.5M-share warrant (strike $20, 5-year term) is exercisable only if Intel’s foundry ownership drops below 51% — a poison pill against a foundry breakup, not a growth-funding mechanism. No board seat; Commerce votes with the Board except narrow carve-outs. (SEC 8-K)
• Nvidia invests $5B and signs on to co-design chips — SPA executed 2025-09-15, announced 09-18, closed 12-26: 214,776,632 shares at $23.28/share. Real technical terms, not just equity: Intel builds Nvidia-custom x86 CPUs for Nvidia’s AI infrastructure and x86 SoCs integrating Nvidia RTX GPU chiplets, connected via NVLink. Intel stock +22% same day, best single-day gain in ~38 years. (Nvidia newsroom)
• Altera divestment closes, Abu Dhabi’s MGX rides along — Silver Lake’s 51% purchase of Altera (announced 2025-04-14, $4.46B cash, ~$8.75-9B implied valuation) closes 2025-09-15; MGX reported to have contributed ~$3.3B of the purchase alongside Silver Lake (09-16/17). (Google News RSS)
• Ohio goes quiet, then “on life support,” then a denied buyout rumor — by 2025-09-05/10 the New Albany site is described as “on life support” three years after groundbreaking, still unfinished, 5+ local leaders gone. Sentiment shifts by mid-2026 as Intel’s stock rallies and the Apple foundry deal raises hopes of an anchor tenant; a 2026-07-21/22 report that SK Hynix was in talks to buy the Ohio campus was denied by SK Hynix the same day, with Intel confirming 07-23 it’s seeking a partner generally. Production timeline has slipped to a reported 2030-2031 range. (Google News RSS)
• 18A stabilizes, 14A finds its first customer, Apple and Microsoft sign on — 18A-P enters risk production 2026-06-16/17 (9% perf gain, 40% lower thermal resistance vs base 18A); by 07-07 the node is reported “finally running smoothly.” Foundry customer wins accumulate through the year: Microsoft’s Maia 2 on 18A/18A-P (~Oct 2025), Tesla’s $20B “Terafab” as first named 14A customer (2026-04-22/23), a preliminary Apple chip-making agreement on 18A-P (announced 05-08, Trump confirms 06-18), and Fortinet on Intel 4 (~07-27/28). No named 18A/14A customer rejection surfaced in this crawl. (Google News RSS)
• CXMT passes Intel’s market cap — 2026-07-26/27: Chinese DRAM maker ChangXin Memory Technologies debuts on Shanghai’s STAR Market, surges ~465-500%, briefly becomes China’s most valuable A-share company; multiple outlets state this surpassed Intel’s own market cap, which was already under pressure (reported ~$79B lost ahead of earnings 07-17, a further ~$90B “wipeout” 07-24). No exact side-by-side dollar figures found — the crossover is confirmed qualitatively, not numerically. (Google News RSS)