Intel

1. Actuals vs Intel's own prior guide (given 2026-04-23)

Revenue growth was +25% YoY ($16.128B vs $12.859B), which Intel's own release calls its "strongest revenue growth in more than fifteen years" [primary, CEO Lip-Bu Tan]. This is a seventh consecutive quarter of revenue above the company's own guidance range (Q1'26 was the sixth, per the pre-print dossier's carried Tan quote) — the streak the pre-print's watch-list flagged as at stake, and it held, by a wide margin.

Consensus context (attributed, not primary): Street sat near ~$14.4B revenue / ~$0.21 non-GAAP EPS heading into the print — the actual $16.128B print beat consensus revenue by ~12%, non-GAAP EPS beat by roughly double.

---

2. The GAAP-vs-non-GAAP divergence — a single, quantified, non-operating driver

This is the one fact that must accompany the headline EPS number, or the print reads backwards. GAAP diluted EPS printed $(2.16), a huge miss against an $0.08 guide and a swing from a GAAP-loss quarter to a much larger GAAP-loss quarter. Non-GAAP EPS printed $0.42, a large beat. Both are true, and the entire gap resolves to one reconciling line:

"(Gains) losses on mark-to-market of Escrowed Shares: $12,529M" [primary: Q2 2026 ER, GAAP-to-non-GAAP net income reconciliation] — "Escrowed Shares refer to shares of Intel common stock held in escrow to be released to the U.S. Department of Commerce (DOC) as we perform and receive cash proceeds in connection with our CHIPS Act Secure Enclave agreement with the U.S. Government." [primary: Q2 2026 ER, cash-flow-statement footnote]

Mechanically: the U.S. government holds an equity-linked position in Intel tied to the CHIPS Act deal. Intel's stock has risen sharply this cycle (pre-print dossier logged +8.64% on 2026-07-21 alone, and one secondary source put INTC +163% YTD into the print ) — and a rising share price increases the mark-to-market value of the shares Intel must eventually deliver, which GAAP books as a non-cash loss to Intel, not a cash outflow. The six-month cash-flow statement shows the same item at $13,619M [primary: Q2 2026 ER, cash-flow reconciliation, H1 basis — the quarterly $12,529M plus a smaller Q1 piece].

Operating income was positive and grew sharply: GAAP operating income $1,796M in Q2'26 vs an operating loss of $(3,176)M in Q2'25 — the core business, ex the Escrowed Shares mark, improved by roughly $5.0B YoY at the operating line. The GAAP net-loss headline is a real accounting number, correctly disclosed, but it is a capital-structure mark on a government instrument, not a demand or execution signal — exactly the kind of fact the non-GAAP reconciliation exists to separate out.

---

3. Foundry — the dollar-loss trend just inverted from the pre-print dossier's own finding

The 2026-07-22 pre-print dossier's own arithmetic (Lens 5A(iv)) flagged, as a load-bearing forensic point, that Foundry's absolute dollar loss grew YoY in Q1 2026 ($(2,320)M → $(2,437)M) even as the margin improved — "only the margin improved... the absolute dollar loss grew." That pattern did not repeat in Q2:

The dollar loss narrowed YoY (−$1,079M, −34.1%) and QoQ (−$348M, −14.3%) — both the margin and the absolute number improved this quarter, which is the first time in this dossier's tracking that both have moved the same direction simultaneously. Foundry revenue itself grew +30.5% YoY (+6.3% QoQ) [primary]. This resolves (partially) watch item #1 from the pre-print list — the dollar trend the desk was watching for reversed, on the good side.

On yield / break-even — still not stated. CFO Zinsner's quote attributes part of the beat to "volume upside driven by higher factory yields and improved cycle times" [primary, CEO/CFO quotes section] — qualitative, no percentage, no die area, no defect density. The strings "yield" appears only in this qualitative form; "break-even" / "breakeven" return zero hits in the earnings release, same as in both filings checked in the pre-print pass. The KeyBanc 85% yield claim remains unconfirmed by Intel itself as of this print — watch item #3 from the pre-print dossier was not answered with a number.

Process/technology — one genuine upgrade over the pre-print framing. The pre-print dossier's best-sourced High-NA claim (from ASML's 6-K) was narrowly scoped: "a process option... on select [18A] product layers... on our existing tool fleet" — explicitly not full HVM. This release states, in Intel's own words:

"Intel Foundry has entered high-volume manufacturing for a subset of Intel® Core™ Ultra Series 3 processors, code-named Panther Lake, using ASML's EXE High NA EUV technology." [primary: Q2 2026 ER, "Foundry Momentum"]

This is a step up from "qualified on select layers" to a named product in actual high-volume manufacturing — the cleanest primary-sourced technical upgrade in the cycle so far. It answers part of watch item #10 (packaging/process disclosure). It is still scoped ("a subset of" Panther Lake) — not a claim that High-NA runs the whole 18A/HVM stack.

Repeated, not new: "18A-P entered risk production" is restated verbatim from the 2026-06-16 newsroom item the pre-print dossier already logged as primary — no incremental information. Fortinet is likewise restated from the 2026-07-21 release already in the pre-print dossier.

External-foundry revenue reconciliation (Altera) — still not answered. This earnings release does not break out a Q2 external third-party foundry + A&T revenue figure or the Altera share of it (that level of detail lived in the 10-Q MD&A/Note 2 in the prior quarter, a document not yet filed for Q2). Watch item #2 remains open pending the Q2 2026 10-Q.

14A pause/discontinue language — not addressed. The earnings release contains no 14A-specific risk conditional; that language lives in the 10-Q risk factors, not the press release, so this print cannot confirm whether it softened or hardened. Not a gap in the print — a gap in what this document type discloses.

---

4. Capex and financing — down, not up, in the actuals; still no FY26 dollar guide

Quarterly gross capex fell YoY, contrary to the "meaningfully increasing investment" framing in the same release:

Against that, CFO Zinsner's on-the-record framing is forward-looking: "to support expected growth this year and next across products and foundry, we are meaningfully increasing our investments in equipment, clean room space, and substrates" [primary, CFO quote] — i.e. the increase is described as coming, not yet visible in the trailing print. Worth tracking whether Q3/Q4 gross capex actually inflects upward, given it has fallen YoY for two straight quarters against that stated intent.

No FY2026 capex-dollar guidance appears anywhere in this release. The only FY2026 dollar figure given is operating expenses (a different line): "Full-Year 2026... GAAP operating expenses approximately $23.0B; Non-GAAP approximately $16.5B" [primary: Q2 2026 ER, supplemental reconciliation table] — this is opex (R&D + MG&A + restructuring), not capital expenditure, and should not be conflated with the ~$17B FY2026 capital plan figure the pre-print dossier carried as web-sourced (TrendForce/wccftech). That capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. figure remains unconfirmed by any primary Intel document seen in either pass.

The financing-structure item that actually moved the cash-flow statement is a partner buyout, not capex spend. "Partner contributions, net" swung to $(12,216)M for the quarter (from +$1,250M in Q2'25) in the Adjusted FCF reconciliation, driving Adjusted FCF to $(8,419)M (from $(1,050)M) [primary]. The six-month cash-flow statement shows the mechanism: "Partner distributions $(14,339)M" vs $(91)M in H1'25 [primary]. This is consistent with — and likely is — the pre-print dossier's already-logged fact that Intel repurchased Apollo's 49% minority stake in the Fab 34 (Ireland) joint-investment entity in Q1 2026: a capital-structure reallocation (buying out a JV partner), not incremental fab spending. Reading the huge negative Adjusted FCF number as "capex is exploding" would be the wrong read — gross capex is down; a one-off partner buyout is what moved the cash line.

The €5B Ireland/Intel 3 investment (announced 2026-07-13, already primary-sourced in the pre-print dossier) is restated in this release's Business Highlights, unchanged — no new dollar figure attached.

---

5. Q3 2026 guidance (new, primary)

[all primary: Q2 2026 ER, "Business Outlook" + supplemental GAAP-to-non-GAAP outlook reconciliation] Consensus context (attributed): the Q3 revenue guide midpoint ($16.3B) sits above a pre-print Street figure reported around $15.06B — guiding meaningfully above where the sell side had been modeling Q3 before this print.

---

6. Market-reaction snapshot (attributed — secondary sources vary on the exact tick)

• Pre-print close: $105.45 on 2026-07-21; INTC had run up sharply into the print (one secondary source cites +163% YTD).
• After-hours reaction: sources disagree on the precise print — one has shares +12.44% to ~$112.70, another describes the move as an initial pop toward $110 that partially reversed intraday-equivalent in after-hours trading. Treat the exact percentage as noisy; the directional read across sources is a clear initial pop with at least partial give-back, not a clean one-way move.
• Options had priced a ~13–15% event move pre-print — the after-hours moves reported sit inside that priced range.

---

Could not verify (this pass)

• 18A yield percentage, defect density, or a Foundry break-even date/figure — still zero company-stated numbers; "higher factory yields" is qualitative only. The KeyBanc 85% claim remains an analyst estimate, not confirmed by Intel.
• External third-party foundry + A&T revenue for Q2, and the Altera share of it — not disclosed at this level of detail in the earnings release; pending the Q2 2026 10-Q.
• Whether the 14A "pause or discontinue" conditional softened or hardened — that language lives in 10-Q risk factors, not the earnings release; pending the Q2 2026 10-Q.
• A confirmed FY2026 capex-dollar target — not present in this release (only FY26 opex is quantified); the ~$17B figure carried from press (TrendForce/wccftech) remains unconfirmed by any primary Intel document across both this pass and the 07-22 pre-print pass.
• Exact after-hours settle price/percentage — secondary sources disagree on the tick; directional reaction (pop, partial give-back) is as far as this pass can confirm without a clean intraday tape pull.