Intel
Integrated device manufacturer: designs and sells x86 client/server CPUs and AI accelerators (DCAI, CCG), plus Altera FPGAs and Mobileye ADAS, while standing up Intel Foundry as a contract manufacturer for external customers (18A/14A). Revenue is product sales + nascent external wafer/packaging fees; economics dominated by fab capex and a deeply loss-making foundry ramp.
- 2026-08-04This market capitalisation previously read ~$667B (as of 25 Jun 2026). Restated to ~$459B on this refresh, roughly 31% lower.
- 2026-08-04This share price previously read ~$132 (as of 25 Jun 2026). Restated to $91.00 on this refresh, roughly 31% lower.
Sources — 21 figures with citations
- Q2 FY2026 revenuefiled2026-06-27 (quarter end); released 2026-07-23$16,128M, up 25% YoY from $12,859Msec.gov — Form 8-K Exhibit 99.1 earnings release; consolidated condensed statements of operations. Derived growth: 16,128/12,859 = +25.4%.
- Q2 FY2026 gross and operating marginsfiled2026-06-27GAAP gross margin 40.4% (vs 27.5%); non-GAAP gross margin 41.8% (vs 29.7%); GAAP operating income $1,796M = 11.1% margin (vs $(3,176)M / (24.7)%); non-GAAP operating income $2,770M = 17.2% (vs (3.9)%)sec.gov — Gross profit $6,509M and operating expenses $4,713M on net revenue $16,128M; non-GAAP gross profit $6,737M. GAAP operating income turning positive is the key change versus the (23.1)% Q1 operating margin.
- Q2 FY2026 bottom linefiled2026-06-27GAAP net loss attributable to Intel $(11,033)M; GAAP diluted EPS $(2.16); non-GAAP net income attributable to Intel $2,197M; non-GAAP diluted EPS $0.42; 5,104M weighted-average diluted sharessec.gov — The GAAP-to-non-GAAP bridge is dominated by a single line: '(Gains) losses on mark-to-market of Escrowed Shares $12,529M' = $2.45 per share.
- Source of the GAAP loss - non-operating chargefiled2026-06-27Interest and other, net $(12,576)M in Q2 FY2026 (vs $(95)M a year earlier), including a $(12,529)M net loss on the change in fair value of the Escrowed Shares derivative liabilitysec.gov — Form 10-Q for the quarter ended 2026-06-27: 'In Q2 2026, interest and other, net included a $12.5 billion net loss from the change in fair value of the derivative liability for the Escrowed Shares.' Bridge check (derived): operating income $1,796M + equity-investment losses $(39)M + interest and other $(12,576)M = loss before taxes $(10,819)M, as reported.
- Escrowed Shares derivative liability (US Department of Commerce)filed2026-06-27Fair value $15,570M at 2026-06-27 vs $2,654M at 2025-12-27; 143M Escrowed Shares not yet released (71M not contingently issuable, 71M contingently issuable); 7M released in Q2 and 13M in H1 2026; YTD mark-to-market loss $13,619Msec.gov — 10-Q Note 4. Shares escrowed under the CHIPS Act Secure Enclave agreement entered with the DOC on 2025-08-22, released as Intel performs and receives cash proceeds. Recognised as $8.8B in other accrued liabilities and $6.8B in other long-term liabilities. Filing states that if the shares are not released by the end of the performance period, half go to the DOC with no consideration and half are forfeited and cancelled. The liability rises with the share price, so a stronger stock mechanically enlarges the GAAP charge.
- Q2 FY2026 operating segment resultsfiled2026-06-27CCPG revenue $8,877M / operating income $2,343M; DCAI $6,262M / $2,474M; Total Intel Products $15,139M / $4,817M; Intel Foundry $5,765M / $(2,089)M; All Other $701M / $230M; Corporate Unallocated $(1,416)M; intersegment eliminations revenue $(5,477)M / $254M; consolidated $16,128M / $1,796Msec.gov — Supplemental Operating Segment Results in the 8-K exhibit. Prior-year comparatives: DCAI $3,939M revenue / $633M operating income; Intel Foundry $4,417M / $(3,168)M.
- Intel Foundry EXTERNAL (third-party) revenuefiled2026-06-27$293M in Q2 FY2026 and $467M in H1 2026, versus $22M in Q2 FY2025 and $53M in H1 2025sec.gov — 10-Q segment-basis-of-presentation note: Intel Foundry segment revenue 'consists substantially of intersegment product and services revenue', plus 'certain third-party foundry and assembly and test revenues from external customers that totaled $293 million in the three months ended June 27, 2026'. Derived: $293M is 1.8% of consolidated revenue and 5.1% of the segment's $5,765M.
- Balance sheet - liquidity and debtfiled2026-06-27Cash and cash equivalents $12,874M + short-term investments $16,853M = total cash and short-term investments $29,727M; total debt $50,537M (short-term $1,988M + long-term $48,549M). Prior year-end: $37,416M cash/STI and $46,585M total debt.sec.gov — 10-Q Liquidity and Capital Resources table, corroborated by the balance sheet in the 8-K exhibit. Also: total assets $202,439M; total Intel stockholders' equity $87,542M; non-controlling interests $15,601M; 5,043M shares issued and outstanding.
- Net debtderived2026-06-27~$20,810M at 2026-06-27, up from ~$9,169M at 2025-12-27sec.gov — Arithmetic: $50,537M total debt - $29,727M total cash and short-term investments = $20,810M. Prior period: $46,585M - $37,416M = $9,169M. Both input pairs are from the same filed liquidity table, so the comparison is like-for-like. Excludes the $15,570M Escrowed Shares derivative liability, which is not debt but is a real claim.
- Q2 FY2026 cash flow and capital expenditurefiled2026-06-27GAAP operating cash flow $7,006M (vs $2,050M); gross capital expenditures (additions to PP&E) $(2,652)M (vs $(4,492)M); proceeds from capital-related government incentives $60M; partner contributions, net $(12,216)M; payments on finance leases $(617)M; adjusted free cash flow $(8,419)M (vs $(1,050)M)sec.gov — Adjusted-FCF reconciliation table in the 8-K exhibit, three months ended. The $(12,216)M net partner contributions line is the Apollo/Ireland SCIP buyout, not recurring operations.
- Underlying free cash flow excluding the Apollo buyoutderived2026-06-27+$3,797M in Q2 FY2026 = +23.5% of revenuesec.gov — Arithmetic: $7,006M operating cash flow - $2,652M gross capex + $60M government incentives - $617M finance-lease payments = $3,797M, i.e. the company's own adjusted-FCF formula with the $(12,216)M net partner-contribution line removed. $3,797M / $16,128M revenue = 23.5%. Reported adjusted FCF margin including it is (52.2)% (derived: -8,419/16,128).
- Capex intensityderived2026-06-2716.4% of revenue in Q2 FY2026sec.gov — Arithmetic: $2,652M gross capital expenditures / $16,128M revenue = 16.44%. Caveat: this is the investing-activities line only. The six-month cash-flow statement also shows $1,423M of 'Additions to property, plant, and equipment' inside FINANCING activities (Arizona SCIP), so total PP&E spend is higher than the investing line implies and this ratio is a floor.
- Ireland SCIP - Apollo minority buyoutfiled2026-06-27Reacquired Apollo's 49% minority ownership interest on 2026-04-08 for $14.2B cash inclusive of transaction costs; eliminated $142M of non-controlling interest, extinguished a $532M derivative liability, and recognised $13.5B as a reduction to capital in excess of par value. Funded with cash, short-term investments and a $6.5B 364-day term loan at 4.79% drawn 2026-04-01 and repaid in full on 2026-04-30 from $6.5B of new senior notes.sec.gov — 10-Q Note 3 (Non-Controlling Interests) and Note 11 (Borrowings). The new notes: $1.0B 4.65% due 2031, $1.0B 5.00% due 2033, $2.2B 5.30% due 2036, $1.8B 6.13% due 2056, $500M 6.20% due 2066. Explains both the $(12,216)M net partner-contribution cash outflow and the drop in common stock and capital in excess of par from $65,185M to $53,780M.
- Q3 FY2026 guidancefiled2026-07-23Revenue $15.8-16.8B; GAAP gross margin 41.0% / non-GAAP 42.0%; GAAP tax rate 1% / non-GAAP 11%; GAAP diluted EPS $0.31 / non-GAAP $0.38sec.gov — Business Outlook section of the 8-K exhibit; gross-margin and EPS outlooks are based on the midpoint of the revenue range. Intel is guiding to GAAP profitability.
- Q2 FY2026 guidance (the beat/miss baseline)filed2026-04-23Guided revenue $13.8-14.8B; GAAP gross margin 37.5% / non-GAAP 39.0%; GAAP diluted EPS $0.08 / non-GAAP $0.20sec.gov — Q1 FY2026 8-K exhibit, Business Outlook. Actual $16,128M revenue was ~$1.3B above the high end; non-GAAP EPS $0.42 was 2.1x the guide; GAAP EPS $(2.16) missed the $0.08 guide by $2.24, all of it the $2.45/share escrowed-shares mark. Q1 FY2026 itself: revenue $13.6B (+7%), GAAP gross margin 39.4% / non-GAAP 41.0%, GAAP operating margin (23.1)%, GAAP EPS $(0.73) / non-GAAP $0.29.
- FY2026 and FY2027 capital expenditure guidance, and 14A customer statusfiled2026-07-232026 capex raised to 'more than $20 billion' (up ~$3B from prior guidance), with tooling spend alone up 40% versus 2025; 2027 capex 'significantly above the 2026 levels' with no figure given. No named or committed external Intel 14A customers disclosed; management pointed to PDK 0.9 delivery scheduled October 2026. Foundry operating loss improved $348M sequentially with no break-even timeline offered. Industry-wide supply constraints across wafers, memory and substrates cited as the dominant challenge.investing.com — Statements by CFO David Zinsner and CEO Lip-Bu Tan on the Q2 FY2026 earnings call. Call commentary, NOT in the filed 8-K or 10-Q. The $348M sequential foundry improvement implies a Q1 FY2026 foundry operating loss of ~$(2,437)M (derived).
- Headcountfiled2026-06-2782.3k total employees at 2026-06-27 (77.6k Intel, 4.7k Mobileye and other subsidiaries), versus 83.2k at 2026-03-28 and 101.4k at 2025-06-28sec.gov — 'Other information' table in the 8-K exhibit. The YoY drop partly reflects the deconsolidation of Altera (~3.0k employees) on 2025-09-12, so it is not all restructuring.
- Share pricemarket2026-08-03$91.00 close on 2026-08-03 (+$0.80 / +0.89%); 52-week range $19.35-$142.35stockanalysis.com — Official close (4:00 PM EDT 2026-08-03), not an intraday level - the same session's intraday high was $91.68 per the history table. The close is ~36% below the 52-week high (derived: 91.00/142.35 - 1 = -36.1%).
- Market capitalizationmarket2026-08-03~$459.0B at the 2026-08-03 closestockanalysis.com — Reported as $459.00B; derivation cross-check using the filed share count: 5,043M shares issued and outstanding at 2026-06-27 x $91.00 = $459.0B. Down from the ~$667B carried in the prior (June 2026) vintage.
- Post-earnings price reactionmarket2026-08-03Closed $100.23 on 2026-07-23 (results day) then $92.32 on 2026-07-24, a 7.9% single-session decline; subsequently $81.88 (2026-07-29) before recovering to $91.00 (2026-08-03)stockanalysis.com — Daily closes. Relevant because the market de-rated the stock into a large revenue and non-GAAP EPS beat - evidence that the equity is priced on the unproven foundry rather than the current P&L.
- Valuation and trailing revenuemarket2026-08-03Forward P/E 54.64x; trailing P/E not applicable (GAAP loss); revenue TTM $57.03B (+7.5%)stockanalysis.com — Data-provider figures at the 2026-08-03 close. The TTM revenue is a provider aggregation - only Q1 FY2026 ($13.6B) and Q2 FY2026 ($16,128M) were independently verified from filings in this refresh.
The thesis on this name
State of AI Compute
Owns two genuine tech firsts (18A PowerVia in production, thick-core glass) and is the only Western second-source optionality on 14A — but the equity is a foundry-turnaround bet swamped by loss-making economics and a poor execution base rate; avoid as the glass/backside-power/foundry expression. The single H2-26→H1-27 14A anchor-customer decision is a sized…
Earnings, margins, COGS & capex
Intel returned to revenue growth in Q1 FY26 (+7% YoY to $13.6B) on a 22% DCAI surge and resilient AI-PC client demand, and posted a record non-GAAP gross-margin beat (41.0% vs 34.5% guide) — its sixth straight expectations beat (estimate: street had ~$12.2B). But the GAAP picture stays deeply negative: a $(3.1)B operating loss and $(3.7)B net loss driven by Intel Foundry's $(2.4)B operating loss as 18A ramps. FY25 was the weakest revenue year since 2010 (~flat at $52.9B), and free cash flow is negative as ~$15–18B/yr capex funds the leading-edge fab build-out. The equity has re-rated violently on AI optimism and government/Nvidia backing despite the loss-making economics — the central fact (analyst targets $48–96 sit far below the ~$132 price).
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~60¢ is cost of goods and ~39¢ operating expense, leaving ~1¢ of operating profit (~27¢ net).
Revenue trend
Margins
up (+2.5pp YoY)
up (record ~650bps beat vs guide)
negative — foundry ramp drag
negative
negative — capex-consuming
COGS structure
COGS is dominated by fab operating costs: wafer fabrication (materials, EUV/High-NA tool depreciation, fab labor, energy), assembly/test/packaging, and — the swing factor — under-utilization and start-up costs as new nodes (Intel 18A, then 14A) ramp before volume absorbs the fixed cost base. Depreciation on ~$15–18B/yr capex is a structural margin headwind; gross margin therefore swings hard on yield, mix (AI-PC/Xeon richer), and node-ramp timing (Q2 FY26 GM guided down to ~39% non-GAAP on heavier early-18A mix).
Capex
~$4.96B gross capex in Q1 FY26 (fact); FY25 ~$14.6B; FY26 guided ~flat YoY as committed demand pulls in capacity. Funds leading-edge fab capacity (Arizona Fab 52 for 18A high-volume manufacturing, Ohio/Oregon), EUV and High-NA EUV lithography for 14A, and advanced packaging (Foveros/EMIB) — the build that underwrites the foundry thesis and the negative FCF.
Latest earnings
Large beat on operations, large GAAP miss on a non-cash mark. Revenue $16,128M came in ~$1.3B ABOVE the high end of the $13.8-14.8B guide. Non-GAAP EPS $0.42 was more than double the $0.20 guide; non-GAAP gross margin 41.8% vs 39.0% guided; GAAP gross margin 40.4% vs 37.5% guided. GAAP EPS $(2.16) versus a guided $0.08 is a $2.24 miss driven entirely by the $2.45/share escrowed-shares mark-to-market. Management framed Q2 as exceeding guidance on 'volume upside driven by higher factory yields and improved cycle times'. Notably the market sold it: the stock fell 7.9% the following session.
Q3 FY2026: revenue $15.8-16.8B (midpoint roughly flat sequentially); GAAP gross margin 41.0% / non-GAAP 42.0%; GAAP tax rate 1% / non-GAAP 11%; GAAP EPS $0.31 / non-GAAP EPS $0.38 - i.e. Intel guides to GAAP profitability in Q3. On the call, CFO David Zinsner raised 2026 capital expenditure guidance to 'more than $20 billion' and said 2027 capex will be 'significantly above the 2026 levels', citing aggressive tool purchase orders, accelerated clean-room build-outs and securing substrate and memory supply against industry-wide supply constraints.
- Intel Foundry operating loss / external revenue
- $(2,089)M operating loss on $5,765M segment revenue (mostly intersegment); only $293M came from external third-party customers in the quarter ($467M in H1 2026), vs $22M in Q2 2025
- DCAI (Data Center and AI)
- Revenue $6,262M, up 59% YoY; segment operating income $2,474M vs $633M a year earlier - the single biggest swing factor
- Escrowed Shares derivative liability (US Dept of Commerce)
- $15,570M at 2026-06-27 vs $2,654M at 2025-12-27; 143M shares still unreleased; $(12,529)M Q2 mark-to-market loss = $(2.45) per share
- Ireland SCIP buyout
- Reacquired Apollo's 49% minority interest for $14.2B cash on 2026-04-08, funded with cash, short-term investments and a $6.5B term loan later refinanced with $6.5B of senior notes; $13.5B booked as a reduction to capital in excess of par
- Headcount
- 82.3k total (77.6k ex-subsidiaries) at 2026-06-27, down from 101.4k a year earlier
- Valuation at the 2026-08-03 close
- 54.64x forward P/E; trailing P/E not meaningful (GAAP loss)
Growth drivers
- DCAI (server CPU) recovery — +22% YoY in Q1 FY26 on Xeon 6/6+ (Clearwater Forest 288-core) and AI-host-CPU demand, the fastest-growing segment
- AI PC client refresh — Panther Lake (Core Ultra Series 3, first 18A consumer chip) with AI-PC >60% of client CPU mix and Windows-refresh tailwind
- Intel Foundry external ramp — 18A wins (Microsoft, Amazon/AWS, US DoD) and 14A pipeline; signed lifetime foundry commitments rose from ~$10B (YE25) to >$15B
- Nvidia partnership — custom Nvidia x86 data-center CPUs (NVLink-coupled) and x86+RTX-chiplet consumer SoCs, a multi-year product + demand driver (early-stage)
- Advanced packaging (Foveros/EMIB) as a differentiated, higher-margin foundry service for chiplet-based designs
- AI inference accelerators — Crescent Island GPU (customer trials 2H26) and Jaguar Shores rack-scale (2027), an inference-first re-entry after Gaudi/Falcon Shores stumbled (early, unproven)
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-01-23. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
The operating turnaround stopped being a promise this quarter. Revenue grew 25% - the fastest in over fifteen years - GAAP operating income flipped positive to $1,796M (11.1% margin) from a $(3,176)M loss, DCAI revenue rose 59% with segment operating income nearly quadrupling, and Intel guides GAAP EPS of $0.31 for Q3. Strip the non-cash government-share mark and the business generated $2,197M of non-GAAP net income and $7,006M of operating cash flow.
- The GAAP loss is an accounting artifact of the stock going UP, not an operating failure. Of the $(11,033)M net loss, $(12,529)M is the mark-to-market on the Escrowed Shares derivative owed to the Department of Commerce. Operating income was +$1,796M. This inverts the earlier pattern, where the loss was real and sat in the operating line.
- Margin recovery is broad and ahead of plan: GAAP gross margin 40.4% (+12.9ppts YoY) and non-GAAP 41.8% (+12.1ppts) both beat guidance, attributed by the CFO to higher factory yields and improved cycle times - the operational variables that actually compound.
- DCAI is inflecting hard: revenue +59% to $6,262M with segment operating income of $2,474M against $633M a year ago. Intel Products as a whole earned $4,817M of segment operating income on $15,139M of revenue - a genuinely profitable core funding the foundry build.
- Demand exceeds supply, which is a first for this story. Management cited industry-wide constraints on wafers, memory and substrates and demand that 'continues to outpace our growing supply', and is raising 2026 capex to more than $20B with 2027 'significantly above' - capex driven by orders, not hope.
- Cost structure is permanently lighter: headcount is 82.3k versus 101.4k a year earlier, and GAAP R&D plus MG&A fell 6% YoY even as revenue grew 25%. Operating leverage is now working in Intel's favour.
- Intel guides to GAAP profitability next quarter ($0.31 GAAP EPS on $15.8-16.8B revenue). If delivered, that is the first clean GAAP-profitable quarter of the turnaround and removes the 'perpetually loss-making' objection.
The foundry - the entire reason to own the re-rating - is still not a business: $(2,089)M of quarterly operating loss on $293M of external revenue, and zero named committed 14A customers. Meanwhile the financing has got worse: net debt roughly doubled to ~$20.8B after a $14.2B cash buyout of Apollo, capex is being raised to >$20B for 2026 and higher again in 2027, and a $15.6B derivative liability to the US government grows every time the stock rises. The market's verdict on the beat was a 7.9% decline.
- External foundry revenue is $293M - 1.8% of consolidated revenue and 5.1% of the Intel Foundry segment's own $5,765M. The segment 'grew 31%' almost entirely by selling to Intel. Nine months after the thesis was priced, the third-party business is still a rounding error.
- Zero named, committed external 14A customers. On the call management pointed to positive feedback expected once PDK 0.9 lands in October 2026 - i.e. the validating catalyst has slipped to the right again, and no signed commitment or customer name was disclosed. No filed source restates the '>$15B lifetime commitments' figure.
- Foundry losses are barely narrowing relative to the spend: $(2,089)M this quarter, an improvement of only $348M sequentially, with no break-even timeline offered. At that rate the losses outlive the current capex cycle.
- The balance sheet materially deteriorated: total debt rose to $50,537M while cash and short-term investments fell to $29,727M, taking net debt to ~$20.8B from ~$9.2B at year-end. Intel paid $14.2B cash for a minority stake it already consolidated, funded with a term loan then $6.5B of new notes at 4.65%-6.20% - including 30- and 40-year paper at 6.13% and 6.20%.
- Capex is going the wrong way for cash generation right after the balance sheet weakened: >$20B guided for 2026 (tooling alone up 40% versus 2025) and 'significantly above' in 2027, against a quarter where the investing-line capex was only $2,652M. The reported free-cash-flow trough is ahead, not behind.
- The government stake is a live, unhedged liability, not just optics: the Escrowed Shares derivative marked up from $2,654M to $15,570M in six months, with 143M shares still unreleased. Every dollar of share-price appreciation creates a further GAAP charge, and if the shares are not released half go to the DOC for no consideration.
- Valuation still has no floor under it from earnings: 54.64x forward P/E at $91, with the stock 36% below its 52-week high and down 7.9% on the session after a large revenue and non-GAAP beat. When a beat that good produces that reaction, the market is pricing something other than the quarter.
What it is worth
Sum-of-the-parts / scenario, not earnings multiple (GAAP-loss-making; forward P/E ~125x is not meaningful). Value = profitable products business (CCG+DCAI+Altera+Mobileye, normalized) + an option on Intel Foundry that swings on landing committed external 14A customers, netted against ~$12B net debt and ongoing capex burn.
$20–45 (low-end targets)
14A fails to win external customers, 18A yields/margins disappoint, AI/government narrative deflates and the equity re-rates toward tangible-product earnings power.
$48–96 (consensus median ~$48
average target ~$93–96) — products recover but foundry stays loss-making and unproven; implies meaningful downside from ~$132.
$150–160+ (analyst high end / 31x a ~2030 ~$6 EPS) — requires committed 14A customers, foundry break-even, and the Nvidia partnership shipping at volume.
At ~$132 the equity has re-rated ~250% YTD and trades far above the $48–96 analyst-target range — pricing foundry success that external revenue ($174M) does not yet evidence; this is a foundry-turnaround option, not a value print.
SWOT
Strengths
- Genuine process firsts — Intel 18A with PowerVia (industry-first backside power delivery) in high-volume production — Panther Lake shipped; transistor-leadership claim at CES 2026
- Only Western/US-based leading-edge logic foundry — strategic second-source to TSMC, the basis of US-government, Nvidia and SoftBank backing (~$15.9B equity in 2025)
- Dominant installed x86 base and Xeon server franchise — DCAI re-accelerated +22% YoY (Q1 FY26) with the 288-core Xeon 6+
- Differentiated advanced packaging (Foveros/EMIB) and a record non-GAAP gross-margin beat (41.0%) showing operating leverage when nodes ramp well
Weaknesses
- Deeply loss-making at the GAAP line — $(3.7)B net loss and $(2.4)B Intel Foundry operating loss in Q1 FY26; negative free cash flow
- Poor multi-year execution base rate — repeated node delays, Gaudi/Falcon Shores cancellations, FY25 weakest revenue since 2010
- Near-zero external foundry revenue ($174M Q1 FY26) — the foundry thesis is almost entirely forward-looking and unproven at scale
- Losing CPU share — AMD nearing record server share and ARM crossing ~50% of hyperscaler CPU compute; x86 secularly pressured
- Capital intensity (~$15–18B/yr capex) consumes cash and leaves ~$12B net debt despite large 2025 equity injections
Opportunities
- Land committed external 14A customers (commitments expected 2H26–1H27) — the catalyst that would validate the foundry as a real TSMC alternative
- Nvidia partnership — custom Nvidia x86 data-center CPUs and x86+RTX consumer SoCs — a credible new demand stream from the AI leader
- US re-shoring / CHIPS / national-security tailwind funding domestic leading-edge capacity (Arizona Fab 52, Ohio)
- AI-inference accelerator re-entry (Crescent Island 2H26, Jaguar Shores 2027) targeting the faster-growing inference market
- Margin expansion as 18A/14A yields mature and under-utilization costs roll off
Threats
- TSMC's entrenched process + ecosystem leadership and AMD on TSMC N2 (Venice) — Intel may regain process parity yet still lack customers
- ARM-based hyperscaler silicon structurally eroding the x86 TAM that funds the whole model
- 14A is a make-or-break bet — failure to win external customers strands tens of billions of fab capex
- Valuation risk — ~$132 price vs $48–96 analyst targets implies the re-rating has front-run un-proven foundry economics
- Government as a ~10% shareholder introduces political/strategic-direction risk alongside its support
Moats, dependencies & bottlenecks
Moats
x86 architecture + installed-base lock-in (software/validation ecosystem) Decades of x86 software and server certification create switching costs, but ARM (~50% of hyperscaler compute) and AMD are structurally chipping at it.
A genuine first in backside power delivery and advanced packaging; durable only if 14A also lands and yields hold vs TSMC.
high (policy-dependent) Scarce strategic asset underwritten by ~$15.9B US-gov/Nvidia/SoftBank capital and national-security demand; monetization still unproven.
~$15–18B/yr capex is a barrier to entry, but the same intensity is the cash-burn liability — a double-edged moat versus the asset-lighter TSMC model.
Entrenched OEM relationships (Dell, HP, Lenovo) and 'Intel Inside', partially offset by AMD/ARM design wins.
Dependencies
14A economics depend on High-NA EUV; ASML is the sole supplier of leading-edge lithography — single point of failure for the roadmap.
Intel still outsources significant compute tiles to TSMC; a competitor it must also pay, capping its own foundry-utilization economics until internal nodes absorb the work.
Microsoft, Amazon, Nvidia, others The whole foundry thesis hinges on landing committed external 14A customers (expected 2H26–1H27); currently zero fully committed.
Backing de-risks funding but introduces political-direction risk and ties the equity to policy continuity.
DCAI/CCG revenue rides the PC refresh and hyperscaler AI capex; a cyclical pullback would hit the products that fund the foundry.
Advantages
- First-to-production on backside power delivery (18A PowerVia) and thick-core glass substrate — genuine technology firsts vs TSMC/Samsung
- Only Western leading-edge logic foundry — irreplaceable second-source for US supply-chain resilience
- Differentiated advanced packaging (Foveros/EMIB) for chiplet-era designs, a higher-value foundry service
- Re-accelerating, cash-generative product franchise (DCAI +22%, AI-PC) funding part of the transition
- Unique capital + customer endorsements: US gov, Nvidia and SoftBank equity plus 18A wins at Microsoft/AWS/DoD
- Integrated IDM model lets Intel be its own lead foundry customer (18A products) to de-risk the node before external ramp
Weaknesses
- Loss-making and cash-burning: $(3.7)B GAAP net loss, $(2.4)B foundry operating loss, negative FCF (Q1 FY26)
- Near-zero external foundry revenue ($174M) — the core thesis is unmonetized
- Poor execution track record — repeated node delays and cancelled AI products undermine forward credibility
- Secular x86 share loss to AMD and ARM in both client and server
- High capital intensity (~28–37% of revenue) and ~$12B net debt constrain flexibility
- Equity valuation (~$132, forward P/E ~125x) far above $48–96 analyst targets — priced for an outcome not yet evidenced
Bottlenecks
- 18A/14A yield maturation — gross margin and foundry break-even hinge on yields climbing fast enough to absorb fixed fab cost (Q2 FY26 GM guided down on early-18A mix)
- Landing committed external 14A customers — without them, tens of billions of fab capex is stranded; the single biggest execution gate
- High-NA EUV tool availability and cost from ASML — paces the 14A ramp timeline (full-scale ~2028)
- Free-cash-flow drain — negative adjusted FCF means the build is funded by debt/equity injections, not operations, limiting financial flexibility
- AI-accelerator credibility — Crescent Island/Jaguar Shores must prove Intel can compete in inference after Gaudi/Falcon Shores failures
- Talent and ecosystem (PDK/IP, EDA enablement) needed to make the foundry usable by external fabless customers
Top signals & trends
Top signals
The single most important catalyst; first named committed customer would validate the foundry — currently zero, all evaluation-stage.
Break-even path is the crux; narrowing loss as 18A yields/volume scale would be the proof point.
~7%/month yield improvement (mgmt) supports the margin beat; any stall would hit gross margin directly.
Guided-down on early-18A mix; watch whether GM re-expands H2 as volume absorbs ramp costs.
+22% YoY is encouraging but ARM crossed ~50% hyperscaler compute and AMD nears record server share — secular pressure persists.
Real products/dates would convert a stock endorsement into revenue; still early-stage with no shipping product.
Trends
Drives DCAI (host CPUs for AI servers, +22% YoY) and the inference-GPU opportunity (Crescent Island).
ARM crossed ~50% of hyperscaler CPU compute; structurally shrinks Intel's core server TAM.
Underwrites Intel as the only domestic leading-edge foundry — funding and defense demand (DoD 18A).
AI-PC >60% of client CPU mix; richer ASPs support CCG and gross margin into the Windows refresh.
Raises barriers (helps Intel as a survivor) but also raises the capex and yield bar Intel must clear to compete.
De-risks funding but adds political-direction and shareholder-conflict risk to a ~10%-owned company.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Sole supplier of EUV and High-NA EUV lithography — critical enabler of 18A/14A.
Deposition/etch/process equipment for Intel fabs.
Etch and deposition tools for advanced-node manufacturing.
Process control / inspection metrology critical to yield ramp.
Supplier and rival — Intel outsources some compute tiles to TSMC N-nodes.
EDA tools + IP needed to make Intel Foundry usable by external customers (Cadence: CDNS).
Major OEM for Intel client + Xeon server CPUs.
PC (HPQ) and server (HPE) OEM channel for Intel silicon.
HK-listed; one of the largest PC/server OEMs shipping Intel CPUs.
Confirmed Intel 18A foundry customer (custom silicon) and large Xeon/Azure buyer.
Intel 18A foundry customer (AI fabric chips) and major data-center CPU buyer.
Prospective buyer of custom Intel-built x86 data-center CPUs under the 2025 partnership.
The leading-edge foundry incumbent Intel must out-execute on yield, ecosystem and customers; AMD/Nvidia/Apple all build there. The benchmark for 18A/14A.
Direct x86 rival taking record server (EPYC/Venice on TSMC N2) and growing client share; also competes in data-center GPUs.
Dominates AI accelerators (Intel's Crescent Island/Jaguar Shores chase inference) — but now also a partner/investor and prospective foundry customer.
Architecture licensor enabling hyperscaler custom CPUs (~50% of hyperscaler compute) that erode x86 — a structural, not product-level, threat.
Korea-listed; the other leading-edge foundry chasing external logic customers and advanced packaging — a non-US second-source alternative.
Custom-silicon houses that design hyperscaler accelerators on TSMC — capture AI-compute spend that bypasses Intel's products.