§ docs  ·  INTC  ·  Customer
ticker
INTC
position
short
conviction
4 / 5
analyst
customer-analyst
company
Intel Corporation
generated
2026-05-03

Customer Analysis — Intel Corporation (INTC)

§ 01Executive View

Intel's customer base is the structural weakness in the short. The product business is anchored by three OEM customers (Dell, Lenovo, HP) who buy a declining-volume, share-losing product (x86 client CPUs) into a late-cycle PC market — and a hyperscaler datacenter customer set that is actively diversifying away from x86 toward in-house ARM (Graviton, Axion, Cobalt) and AMD EPYC. The foundry "second customer base" — the entire bull thesis — is not yet a customer base: external foundry revenue ran ~$50M in mid-2025 against a ~$17.8B IFS segment, with one Microsoft chip and a few small DoD RAMP-C engagements as the only marquee names. That is press-release demand, not pull-through demand. In short, Intel's existing customers are eroding faster than its new customers are committing.

§ 02Customer Concentration

Metric Latest (FY2024 10-K basis, 2025 ests) YoY change Source
Top customer % (Dell) ~19% (FY23 disclosed; est. ~17–19% FY24/25) Drifting down — AMD EPYC server share at Dell is rising Intel 10-K FY2023; concentration risk language carried forward in FY2024 10-K
Top 3 (Dell + Lenovo + HP) ~40% (FY23 disclosed) Roughly flat as % of a shrinking client business; absolute volume falling Intel 10-K FY2023
Top 10 Not separately disclosed; inferred ~55–60% Likely rising as hyperscalers consolidate 10-K segment discussion
Named >10% customers Dell, Lenovo, HP (FY23). Status of HP at >10% in FY24/25 unclear — may have dropped below threshold HP softening; if HP slips below 10%, the disclosed customer set narrows to two 10-K FY2023, FY2024
External foundry revenue ~$50M (Q2 2025 10-Q acknowledged) vs $17.83B reported IFS segment IFS segment "revenue" is ~99.7% internal transfer pricing, not external customer cash CFO Zinsner public commentary May 2025; Motley Fool reporting

Concentration interpretation for the short. A 19% Dell, 11% Lenovo, 10% HP triangle is the classic profile of a late-cycle franchise: three customers who all sell into the same end market (consumer + commercial PC), face the same secular headwind (PC unit decline + ARM-on-Windows), and have an active second-source supplier (AMD) they are gradually shifting share to. This is the worst-of-both-worlds concentration profile — it is concentrated enough to make any one customer's procurement decision a needle-mover, but it is not sticky the way a TSMC-Apple relationship is sticky, because there is no qualification moat on x86 client CPUs (OEMs already qualify both Intel and AMD on the same SKUs).

§ 03End-Market Exposure

Intel reorganized segments in Q1 2025 (NEX folded into CCG and DCAI). Approximate FY2025 mix below; demand-quality column is the lens that matters for the short.

Segment % rev Cycle position Structural Macro sensitivity Demand quality
Client Computing (CCG) — PC OEMs ~46% Late-cycle; H1'25 had tariff pre-buy pull-in Declining: PC unit volume flat-to-down secularly; AMD share gains; Snapdragon X (ARM-on-Windows) entering High (consumer discretionary + enterprise refresh) Mixed-to-poor: H1'25 tariff pre-buy was acknowledged channel/inventory build that creates a "demand vacuum" in H2'25–H1'26 per analyst commentary
Datacenter & AI (DCAI) — Xeon CPUs to hyperscalers + enterprise ~24% Mid-cycle but mix-deteriorating Mixed: x86 share at ~64% Mercury Q1'26 vs 77% in 2021; hyperscaler ARM (Graviton, Axion, Cobalt, Vera CPU) carving structural share; AMD EPYC at the high end High (hyperscaler capex cycle) Q1'26 22% YoY growth flattered by AI-server CPU pull-through (1:4 CPU:GPU ratio) — but this is Nvidia and AMD demand pulling Intel CPUs along, not Intel winning customers
Intel Foundry external <0.3% (~$50M of $17.83B IFS revenue) Pre-revenue Aspirational N/A Almost entirely internal transfer; external customer count in the single digits
Mobileye + Other ~5% Late-cycle in legacy ADAS EyeQ ASP declining ~$0.50/yr from China OEM mix; FY25 guide cut 3% below consensus High (auto build rates) Top 3 customers ~53% concentration; Q4'25 -9% YoY
Network & Edge (now distributed into CCG/DCAI) absorbed Late-cycle Telco capex weak; segment was dismantled in 2025 Med The fact that NEX was folded back is itself a customer-base signal

Weighted demand picture. ~70% of Intel's revenue (CCG + DCAI Xeon to non-AI workloads) is exposed to two structurally declining demand curves: PC unit decline + x86 server share loss to ARM and AMD. The Q1'26 DCAI 22% YoY headline is real but reads as a cyclical AI-server CPU pull-through rather than a structural share recovery — Intel still ceded share to AMD over the trailing four quarters per Mercury, and the "CPU:GPU ratio is rising from 1:8 to 1:4" pitch is itself an admission that the customer pulling the CPU is the GPU vendor, not the end buyer. Mobileye is a third declining curve. The new curve (foundry external) has not yet produced a customer base.

§ 04Contract Structure & Switching

Client (CCG): No long-term take-or-pay. OEMs run dual-sourced bills of materials (Intel + AMD on the same SKUs) and shift mix quarterly based on price/performance. Switching cost from Intel → AMD on a given laptop SKU is days-to-weeks of qualification, not years. The ARM-on-Windows transition (Snapdragon X) requires more lift but Microsoft, Lenovo, and HP have already shipped commercial volumes, meaning the qualification cost is now amortized — once paid, future ARM SKUs slot in incrementally. This is the textbook low-switching-cost customer relationship the short thesis needs.

Datacenter (DCAI): Hyperscalers run their own internal qualification stacks. AWS, Google, Microsoft, Meta have all already qualified both AMD EPYC and their own ARM silicon at production scale alongside Intel Xeon — the qualification moat that historically protected Intel's hyperscaler share is already breached. Enterprise IT customers (Dell, HPE, Lenovo as channel) carry more inertia, but enterprise is the slower-growing slice; hyperscaler is where the volume swing happens.

Foundry (IFS external): A foundry customer relationship at the leading edge has real multi-year switching cost — PDK, IP libraries, design qualification, package co-design — typically 18–36 months from tape-out to volume. This is normally a moat, but for Intel it cuts the wrong way: every fabless customer that has not committed to 18A by 2026 is structurally going to TSMC N2 or A16, and the lock-in then runs against Intel for the next node cycle. The named external customer set as of 2025 — Microsoft (one chip), DoD via RAMP-C (Trusted Semiconductor Solutions, Reliable MicroSystems — both small defense-base players, not commercial volume), no Apple, no Nvidia, no AMD, no Qualcomm, no MediaTek, no Broadcom, no Marvell — is a narrow external book against the stated "world's #2 foundry by 2030" ambition. Intel's stated $15B "lifetime deal value" pipeline figure is a marketing number, not a backlog; Intel does not disclose a foundry RPO comparable to TSMC's customer prepayment book.

Backlog visibility: Intel does not report RPO. The closest proxy is Mobileye's lifetime design-win value (which is growing — the one bright spot) and IFS internal-transfer revenue (which is not a customer commitment). Forward visibility on the product side is one to two quarters, normal for the merchant CPU model.

§ 05Demand Quality

This is where the short framing is sharpest. Intel's 2025 customer demand has been visibly low quality across three distinct vectors:

  1. Tariff pre-buy in H1'25 (channel fill, not pull-through). Intel's own Q1 2025 commentary, picked up across IG and analyst coverage, acknowledged that PC OEMs and channel pulled forward orders ahead of tariff uncertainty. CCG H1'25 strength was visibly inventory build, not consumer pull. Analysts widely flagged a corresponding "demand vacuum" in H2'25–H1'26 — and CCG was indeed -4% QoQ in a subsequent quarter despite AI PC unit growth of +16%, suggesting the AI PC narrative is masking ongoing volume erosion in the rest of the client book.
  2. AI-CPU pull-through is GPU-pull, not Intel-pull. Q1'26 DCAI growth of 22% was driven by AI servers needing host CPUs. The customer here is functionally Nvidia (and to a lesser extent AMD) — they are the ones designing the rack architectures that pull Xeon along. If hyperscalers move to ARM-based Vera (Nvidia's own CPU rack) or Graviton-paired-with-Trainium configurations, Intel loses this pull-through entirely. This is demand without a customer relationship — Intel is a passenger in someone else's customer relationship.
  3. Foundry "wins" are press-release-grade, not commitment-grade. Microsoft's 18A engagement is "one chip." RAMP-C is small, cost-plus defense work. The conspicuous absence — Apple, Nvidia, AMD, Qualcomm, MediaTek, Broadcom, Marvell — is the evidence. If 18A's customer pitch were working, at least one of those names would have a public commitment by mid-2026. They don't. CFO Zinsner's own May 2025 admission that committed external volume "is not significant right now" is the cleanest demand-quality signal in the entire memo. The Falcon Shores cancellation in early 2025 — Intel's flagship merchant AI accelerator pulled to "internal test chip" because the Gaudi 3 customer adoption fell short of even a $500M target — is the demand-quality signal on the merchant AI side.

The Gaudi tell. Intel publicly admitted Gaudi 3 "did not meet the $500M target" because of software immaturity. $500M is the floor of what counts as a credible AI accelerator business in 2025 (NVDA datacenter is running ~$30B/quarter, AMD MI is multi-billion). A merchant AI customer base never formed. This same software/ecosystem deficit ("PDK / IP / EDA ecosystem gap vs TSMC" in synthesis.md) is the same structural reason 18A external customers are not committing.

§ 06Bull Points

(Required for completeness — these are the long-side counters the short must defeat.)

  • DCAI Q1'26 +22% YoY beat AMD's data-center growth for the first time in 19 quarters; if this is a genuine inflection, AI-server CPU pull-through is durable through the Rubin/MI450 cycle.
  • Microsoft 18A engagement, while small, is a marquee name. If it produces yield/performance evidence good enough to attract a second hyperscaler in 2026–2027, the foundry customer book can compound.
  • Mobileye design-win lifetime volumes from Q1 2025 alone "represent a high majority of what was achieved all of last year" — the ADAS customer book is rebuilding under EyeQ6L, including new European and Korean OEMs.
  • Concentration risk is symmetric on the long side: a 19% Dell relationship is also a deep, multi-decade qualification footprint that AMD has been chipping away at for years and still hasn't fully captured. Inertia cuts both ways.
  • US government as customer (CHIPS Act, RAMP-C) is a non-economic buyer with strategic reasons to keep Intel viable — provides a floor on foundry external revenue regardless of commercial commitment.

§ 07Bear Points

  • Top-3 OEM concentration (~40%) is in a structurally declining end-market (PC) with active dual-source competition (AMD) and an emerging third source (Snapdragon X ARM).
  • Hyperscaler datacenter customer base has already qualified every alternative (AMD EPYC, ARM Graviton/Axion/Cobalt/Vera) — the customer-relationship moat is structurally broken on the segment that contributes ~24% of revenue.
  • External foundry revenue is ~$50M against $17.83B IFS segment. The "second customer base" that the long thesis requires does not yet exist in commercial volume; CFO has publicly acknowledged this.
  • Named external foundry customers (Microsoft single chip, DoD RAMP-C defense base) do not include a single fabless leader — Apple, Nvidia, AMD, Qualcomm, MediaTek, Broadcom, Marvell — that would be required to make Intel "world's #2 foundry by 2030" credible.
  • H1'25 client revenue strength was acknowledged tariff pre-buy / channel fill — low-quality demand that creates a forward inventory drag, not a pull-through baseline.
  • Gaudi 3 missed even a $500M target and Falcon Shores was cancelled to internal test chip — the merchant AI accelerator customer base never formed, removing the highest-value end-market entirely.
  • Mobileye top-3 concentration ~53% with Q4'25 -9% YoY and ASP eroding from China OEM mix — the auto vertical is also softening.
  • Network & Edge segment was dismantled in 2025 (folded into CCG/DCAI) — a customer base too small to justify standalone reporting is itself a signal.

§ 08Conviction (1–5)

4 of 5. The customer-side evidence is unusually clean for a short: the existing customer base is structurally eroding across three of four end-markets (PC, datacenter x86, ADAS), the new customer base (foundry external) is essentially empty, the merchant AI customer base has been formally abandoned (Falcon Shores), and management itself has publicly acknowledged the foundry external commitment gap. The reason this is not a 5 is the Q1'26 DCAI inflection — if AI-server CPU pull-through proves durable and genuinely structural rather than cyclical Nvidia-pull, the bear case on datacenter softens. The 14A/DSA "magic bullet" question (synthesis.md, contested claim #5) is binary and not in scope here, but a positive 14A surprise late-2027 would invalidate the foundry-customer leg.

§ 09Key Risks to This Read

  1. Q1'26 DCAI inflection durable, not cyclical. If x86 server share stabilizes at ~64% rather than continuing its 2021→2026 slide, the Xeon customer base is more resilient than this memo assumes.
  2. A second hyperscaler signs 18A. Google was rumored as "likely to follow" Microsoft per industry press. If a credible second commercial customer emerges in 2026, the demand-quality argument on foundry weakens materially.
  3. Tariff regime extends. A persistent tariff structure that keeps OEMs structurally over-ordering would convert "channel fill" into a sustained demand floor.
  4. AI PC pull-through. If AI PC really does drive a multi-year client refresh cycle (Nova Lake, Panther Lake) at premium ASPs, CCG could grow even with declining unit volume.
  5. CHIPS Act / national security buyer. US government could anchor IFS at non-commercial economics for a decade — moves the foundry customer count question from "can it win Apple" to "does it need to."
  6. Concentration cuts both ways on the long side. The deep Dell/Lenovo/HP qualification footprint is itself a moat that has resisted AMD for two decades.

§ 10Sources

  1. Intel Corporation 10-K FY2023 (filed Jan 2024) — disclosed top-3 customer percentages: Dell ~19%, Lenovo ~11%, HP ~10%. https://www.intc.com/filings-reports/all-sec-filings/content/0000050863-25-000009/0000050863-25-000009.pdf (FY2024 10-K filed 01/31/2025 carries forward the concentration-risk language.)
  2. Intel Q1, Q2, Q3, Q4 2025 financial results press releases (intc.com investor relations) — segment revenue and CCG/DCAI commentary.
  3. Intel Q1 2026 financial results — DCAI +22% YoY, $13.6B revenue. https://www.intc.com/news-events/press-releases/detail/1767/intel-reports-first-quarter-2026-financial-results
  4. Intel Q4 2025 earnings call transcript (The Motley Fool, Jan 23 2026). https://www.fool.com/earnings/call-transcripts/2026/01/23/intel-intc-q4-2025-earnings-call-transcript/
  5. Intel Q1 2026 earnings call transcript (The Motley Fool, Apr 23 2026). https://www.fool.com/earnings/call-transcripts/2026/04/23/intel-intc-q1-2026-earnings-transcript/
  6. Stock Analysis "Intel Revenue by Segment" — FY2025 segment mix (CCG 45.69%, DCAI 23.99%, IFS 25.27%, Other 5.05%). https://stockanalysis.com/stocks/intc/metrics/revenue-by-segment/
  7. Mercury Research server CPU report Q1 2026 — Intel x86 server share 64.2% vs 77% in early 2021 (cited via 24/7 Wall St.). https://247wallst.com/investing/2026/04/28/arm-holdings-amd-intel-crater-in-cpu-stock-shock/
  8. Motley Fool, "Intel Foundry Is Still Struggling to Win Customers" (May 15 2025) — CFO Zinsner acknowledgement that committed external 18A volume is "not significant right now"; ~$50M external foundry revenue in Q2 2025 10-Q. https://www.fool.com/investing/2025/05/15/intel-foundry-is-still-struggling-to-win-customers/
  9. Intel Newsroom, "Intel Foundry Adds New Customers to RAMP-C Project for US Defense" — Trusted Semiconductor Solutions and Reliable MicroSystems added to RAMP-C (Jan 2025). https://newsroom.intel.com/intel-foundry/intel-foundry-adds-customers-ramp-c-project-us-defense
  10. Intel Newsroom, "Intel Foundry Achieves Major Milestones" — 18A milestones and Microsoft engagement context. https://newsroom.intel.com/intel-foundry/intel-foundry-achieves-major-milestones
  11. Fortune / Yahoo, "Intel's AI dreams slip further out of reach as it cancels its big data-center GPU hope, Falcon Shores" (Jan 2025) — Falcon Shores cancelled to internal test chip; Gaudi 3 missed $500M sales target. https://fortune.com/2025/01/31/intels-ai-dreams-slip-further-out-of-reach-as-it-cancels-its-big-data-center-gpu-hope-falcon-shores/
  12. TechCrunch, "Intel won't bring its Falcon Shores AI chip to market" (Jan 30 2025). https://techcrunch.com/2025/01/30/intel-wont-bring-its-falcon-shores-ai-chip-to-market/
  13. SDxCentral, "Intel dismantles network and edge unit" — NEX absorbed into CCG/DCAI in Q1 2025. https://www.sdxcentral.com/news/intel-drifts-further-from-networking-as-it-dismantles-nex-unit/
  14. IG International, "Intel Q1 2025 Earnings Preview: Tariff Concerns Boost PC Sales" — H1'25 tariff pre-buy / channel-fill commentary. https://www.ig.com/en/news-and-trade-ideas/Intel-Q1-2025-earnings-preview-what-to-expect-as-tariff-fears-drive-temporary-demand-surge-250417-250417
  15. Mobileye Q3 2025 results press release — revenue, customer concentration, EyeQ6L design wins. https://ir.mobileye.com/news-releases/news-release-details/mobileye-releases-third-quarter-2025-results-and-provides
  16. Mobileye Q1 2025 results press release. https://ir.mobileye.com/news-releases/news-release-details/mobileye-releases-first-quarter-2025-results-and-reaffirms-full
  17. Tom's Hardware, "We can't completely vacate the client market says Intel amid wafer supply shortages" — Nova Lake / Panther Lake / 14A timeline context. https://www.tomshardware.com/pc-components/cpus/we-cant-completely-vacate-the-client-market-says-intel-amid-wafer-supply-shortages-nova-lake-still-on-track-for-late-2026-release-14a-in-2028
  18. Internal corpus: synthesis.md (cohort domain synthesis, 2026-05-03) — Intel sentiment -1, "Embattled foundry; 14A is binary"; PDK/IP/EDA ecosystem gap as named risk; foundry external customer set framing.
  19. Internal corpus: companies.json entry id 39 (Intel) — supporting quotes including SemiAnalysis "Intel on the Brink of Death" reference, BSPDN as plausible foundry asset, conservative post-Gelsinger leadership pulling back.

Works cited

  1. Intel Corporation Form 10-K for fiscal year 2024
    filing first cited by · macro-analyst
    • Revenue geography mix, segment revenue split (CCG/DCAI/NEX/IFS), long-term debt schedule, fab footprint (US, Israel, Ireland), capex commitments, hedging program disclosure
  2. Intel Corporation Form 10-Q filings, quarters of fiscal 2025
    filing first cited by · macro-analyst
    • Quarterly debt waterfall and refi schedule, segment revenue trend, capex commentary, net interest expense trajectory
  3. Intel investor day materials and Foundry Direct Connect 2024/2025 disclosures
    investor_materials first cited by · macro-analyst
    • IFS strategy, 18A and 14A node timing, external customer disclosure (Microsoft, DoD), capex commitments
  4. BLS industrial wage and energy/utility inflation series
    macro_data first cited by · macro-analyst
    • Input-cost inflation framing for Intel's US-domestic opex base
  5. ICE US Dollar Index (DXY) history, 2021–2026
    market_data first cited by · macro-analyst
    • USD-strength regime characterization and FX headwind sizing
  6. Reporting on Intel Kiryat Gat campus expansion and regional security environment (2023–2025)
    public_reporting first cited by · macro-analyst
    • Israel single-country geopolitical tail framing for fab base and opex
  7. US Treasury constant maturity yield series, 2021–2026
    market_data first cited by · macro-analyst
    • Rate-regime characterization and rate-sensitivity beta context
  8. Companies registry — INTC entry
    internal first cited by · macro-analyst
    • Intel sentiment, role, mention-count, supporting quotes, catalysts, risks, and contested-claim flags
  9. Semiconductor cohort synthesis
    internal first cited by · macro-analyst
    • Cohort-level macro and geopolitical framing for INTC
    • AI capex super-cycle context Intel is under-exposed to
    • foundry binary framing