No NEW named external 18A/14A customer at Intel's Q2 2026 print; Fortinet (Intel 4) is the only fresh name; >$20B capex is management-asserted commitment-gated but the disclosed driver is internal Xeon/DCAI demand, not confirmed external leading-edge POs.

Autoresearch: Intel Foundry (INTC) external 18A/18A-P/14A customers around the Q2 2026 print

Generated by /autoresearch on 2026-07-24. Synthesized across 3 rounds from ~8 web pages/searches; no Grokipedia anchor (no entry — HTTP 404 on candidate slugs). Several primary sources (CNBC, Seeking Alpha, Tom's Hardware, techtimes) returned 403 and were routed around via reliable mirrors (Motley Fool, StockTitan, BigGo, TradingView, Intel/AWS newsrooms). See Provenance. Treat as raw material — review before promoting. Context: none (target: stock-market; INTC foundry theses)

Summary

No new named external customer for Intel's leading-edge nodes (18A / 18A-P / 14A) materialized at or around the Q2 2026 earnings print (July 23, 2026). The only fresh named external customer is Fortinet (disclosed July 21), and it is on Intel 4 — a mature node, a security-ASIC (SP6) win, explicitly not 18A/14A (Motley Fool, CNBC via search). The pre-existing named 18A external customers — AWS (AI fabric chip, announced Sept 2024) and Microsoft (unnamed custom chip, 2024) — were not refreshed with new commitments at this print (Intel Newsroom, Manufacturing Dive). On 14A, Intel still describes only two unnamed prospective customers, with firm supplier decisions expected H2 2026 into H1 2027 (Tom's Hardware via search). Apple remains an evaluator of 18A-P (PDK received, simulations run, low-end/legacy chips rumored) with no formal commitment (CNBC/MacDailyNews).

Capex-vs-PO signal (the tradeable core): Intel raised FY2026 capex from ~$18B to >$20B and guided FY2027 "meaningfully higher" (Seeking Alpha via search). Management (Zinsner) asserted the increase is gated by "concrete customer commitments," "long-term agreements," and "customer prepays," and that they are "aggressively locking in tool purchase orders" (BigGo call transcript, BigGo recap). But the disclosed, verifiable demand driver is Intel's own products — record DCAI/server growth (+59% YoY on Xeon 6), AI businesses ~70% of revenue — not confirmed external leading-edge purchase orders. External foundry revenue was only ~$293M (against a $5.8B total Foundry segment that is overwhelmingly internal), so >$20B capex is running far ahead of any disclosed external commitment. Net: the capex raise is best read as internal-demand-led capacity + management-asserted (but unnamed, unquantified) foundry LTAs, not evidence of a signed marquee external 18A/14A order.

Findings

1. Named external customers: Fortinet is new (but Intel 4), nothing new on 18A/14A

Fortinet is Intel Foundry's first named external customer under CEO Lip-Bu Tan, announced ~July 21, 2026 — an SP6 security chip on the Intel 4 process, not 18A/14A (Motley Fool). The Motley Fool framing is explicit: "A named customer on Intel 4 is progress. It isn't the marquee win that would prove Intel can manufacture the industry's most advanced chips in large quantities," and Intel "has yet to land a major one for its most advanced manufacturing processes, known as 14A and 18A."
Intel's own Q2 2026 press release references only a "strategic collaboration with Fortinet" and names no external customer for the 18A family; it notes 18A-P "entered risk production, meeting the timeline shared with customers and partners last year" (StockTitan / Intel PR summary).
The Q2 2026 earnings-call transcript names no external foundry customer for 18A/18A-P/14A. Tan spoke to "positive customer feedback" and excitement once customers "see the 0.9 PDK, the yield," but disclosed no commitments (BigGo transcript).

2. Pre-existing 18A names (AWS, Microsoft) — not refreshed at this print

AWS: Sept 16, 2024 multi-year, multi-billion-dollar co-investment — Intel to produce an AI fabric chip on 18A (plus a custom Xeon 6 on Intel 3) (Intel Newsroom, Electronics Weekly). This is the "lighthouse" 18A external win — but it dates to 2024, not the July 2026 print.
Microsoft: 2024 pact to manufacture an unnamed custom chip on 18A (Manufacturing Dive). Described as a "design partner." No new 2026 commitment surfaced at the print.
Caution: one search-engine summary asserted a fresh "order confirmation from a cloud CSP" with 85% 18A yield, and an April 2026 FinancialContent piece framed an "AWS 18A deal" as a current catalyst — but that summary also cited a stale/incorrect "$14.4B" revenue figure (actual Q2 was $16.1B), so the "new CSP order" claim appears to be a conflation of the 2024 AWS deal, not a print-day event. Treated as unverified (see Contradictions).

3. 14A: two unnamed prospects, decisions H2 2026 → H1 2027; internal risk production H2 2027

Intel maintains it has two prospective external 14A customers and expects firm supplier decisions "starting in the second half of 2026, extending into the first half of 2027" (Tom's Hardware via search). No names.
14A remains on track for risk production for internal products in H2 2027, with a Q2-2026 decision to fully commit to a high-volume ramp in 2028 (BigGo recap).

4. Apple 18A-P: closer, still no commitment

18A-P entered risk production (June 2026); Apple reportedly received the PDK and ran internal simulations that "met expectations," with rumored low-end/legacy iPhone/iPad/Mac chips as the entry vehicle — but no formal commitment (MacDailyNews, MacDailyNews test report). Analyst commentary frames Apple as a potential "once-in-a-generation lifeline" but notes ~80% of the hypothesized volume hinges on a single iPhone chip (Wccftech).
No print-day evidence of Nvidia, AMD, or Broadcom committing 18A/14A wafers in any fetched source.

5. The capex-vs-PO read (tradeable implication for INTC)

The numbers: Q2 2026 revenue $16.1B (+25% YoY), non-GAAP EPS $0.42 (vs $0.20 guide), GAAP EPS $(2.16); Intel Foundry segment revenue $5.8B (+31% YoY) with external revenue ~$293M; capex raised to >$20B for 2026 (from ~$18B), 2027 "meaningfully" higher (StockTitan, Seeking Alpha via search, search recaps).
Management's framing (commitment-gated): Zinsner — capex is "gated by concrete customer commitments," reflects "customers with long-term agreements," aided by "customer prepays," and Intel is "aggressively locking in tool purchase orders" while remaining "very disciplined"; capex deployed only "when we feel very confident we can generate a very good return" (BigGo transcript, BigGo recap).
The skeptical read (what's actually disclosed): The verifiable demand is internal — record DCAI/server +59% YoY (Xeon 6), AI ~70% of total revenue — i.e., Intel building capacity for its own product resurgence. External foundry revenue is de-minimis ($293M) and no external leading-edge customer is named or quantified. So >$20B capex is running well ahead of any disclosed, external purchase-order book; the "customer commitments" backing the raise are unnamed and unverifiable. TradingView's recap notably framed the raise as capacity "responding to market conditions rather than locked customer commitments" (TradingView) — a direct tension with the call's own language.
Tradeable takeaway: The bull case (foundry LTAs unlocking capacity; Apple/14A commitments due H2 2026) is real but still promissory. The print delivered an internal-execution beat (Xeon/DCAI + 18A yield/output up ~50% QoQ) rather than the external-customer proof point. A high-conviction "external 18A/14A marquee customer" thesis is not yet confirmed by this print; the next binary catalyst is the H2 2026 → H1 2027 window for 14A firm decisions and any Apple 18A-P commitment. Watch external foundry revenue trajectory as the falsifier — if capex keeps rising while external revenue stays ~$0.3B/qtr, the "speculative build" reading strengthens.

Contradictions and open questions

"New CSP 18A order at the print" — unverified/likely conflated. A search summary claimed an 85%-yield 18A "order confirmation from a cloud CSP," but tied it to a stale $14.4B revenue figure and an April-2026 AWS framing. No fetched primary source corroborates a new CSP order at the July 23 print. Likely a re-surfacing of the Sept-2024 AWS 18A deal. Verify against the full transcript before treating as a fresh catalyst.
Capex characterization conflict. The earnings call (BigGo) frames the >$20B raise as commitment/LTA-gated; TradingView's recap frames it as demand-responsive capacity without locked commitments. Both can be true (internal demand + unnamed foundry LTAs), but the external purchase-order backing remains undisclosed and unquantified.
External foundry revenue figure. "$293M external" comes from search recaps of the call; Intel's press release did not separately break out external foundry revenue. Confirm the exact figure and its QoQ trend from the 10-Q / call.
Apple timing. "Closer" but uncommitted; the H2-2026 commitment window Tan floated in May is the key unresolved catalyst. No firm date.
Several strong primary sources (CNBC, Seeking Alpha, Tom's Hardware, techtimes) 403'd on direct fetch — findings on those points rest on reliable mirrors/recaps (Motley Fool, StockTitan, BigGo, Intel/AWS newsrooms) and should be confirmed against the primary transcript when accessible.

Provenance

Rounds run: 3 (full)

Sub-questions by round:

Round 1 (broad survey):

What did Intel's Q2 2026 print (July 23) say about 18A external customers/revenue?
Is there any named external 18A/14A customer beyond internal products?
What are the raised FY2026/FY2027 capex figures and rationale?

Round 2 (drill-down):

Is the "confirmed 18A deal with a major cloud service provider" a new print-day event? — targeted the highest-stakes named-customer claim
Did Zinsner/Tan tie the >$20B capex to committed POs vs speculative build? — targeted the capex-vs-PO core
Apple 18A-P status around the print — targeted the leading external-customer candidate

Round 3 (resolve remaining uncertainty):

Intel's own press release + call transcript on external customers and capex — targeted primary-source confirmation
Are AWS/Microsoft pre-existing (2024) 18A names or new? — targeted separating old wins from print-day news

Anchor source (Grokipedia): no entry found — HTTP 404 on candidate slug Intel_Foundry_Services. Proceeded without an anchor.

URLs fetched (4 successful, 3 failed / routed around):

Round 1:

Motley Fool — Intel's foundry lands first named outside customer — news/analysis — confirmed Fortinet = Intel 4, no 18A/14A named customer.
~~CNBC — Intel Foundry lands Fortinet~~ — HTTP 403 (content recovered via search snippet + Motley Fool).
~~Seeking Alpha — Intel Q3 guide + 2026 capex >$20B~~ — HTTP 403 (capex figure recovered via search snippet).
~~Tom's Hardware — two 14A prospective customers~~ — header-only / soft-paywall (headline claim recovered via search).

Round 2:

TradingView — key facts: 2026 capex >$20B, Q3 +$2.6B, 42% GM — news recap — capex framed as demand-responsive; China CPU deals; no named foundry customer.
~~techtimes — Intel Foundry breaks silence with Fortinet~~ — HTTP 403.

Round 3:

StockTitan — Intel Q2 2026 financial press-release summary — primary PR mirror — $16.1B rev, Foundry $5.8B, no named external customer, Fortinet collaboration only.
BigGo — INTC Q2 FY2026 earnings call transcript — call transcript — no named external customer; capex "gated by customer commitments / long-term agreements / prepays," but no names.
BigGo — Q2 2026 earnings recap — recap — 14A internal risk production H2 2027, ramp 2028; external revenue $293M.

Supporting search-surfaced sources (not directly fetched):

Tools used: WebSearch, WebFetch, grokipedia-fetch (skill — no entry). Generated: 2026-07-24