PM thesis · INTC · Intel Corporation · 2026-05-03 · cohort architecture role: SHORT

INTC

Intel Corporation

Short

Conviction: 3 / 5
Sizing: Small · probe
Horizon: 9–18 months
Entry: 1/3 · 1/3 · 1/3

A structurally impaired IDM whose product franchise has lost share in every segment for five consecutive years, whose financials show negative ROIC against a 12.7% WACC and −$48B cumulative FCF over four years, trading at ~22× forward / ~38× EV/EBITDA after a 129% YTD rally — a textbook "narrative crowded into a still-broken P&L" setup. The reason this is a small probe rather than a medium short: ~35% probability of federal equity injection and ~50% probability of CHIPS covenant loosening flatten the asymmetry to ~1:1.3. Pair-trade with TSM long is the preferred expression.

Underlying analysis

PM thesis competitor supply chain customer financial market regulatory macro sources

§ 01

Asymmetry · ~1 : 1.3 · the binding constraint.

does not clear the 2:1 short bar

The asymmetry test is the binding constraint. Five of seven analysts bearish at conviction 4 — but the regulatory cushioning leg makes the timing-honest payoff symmetric, not convex. The pair-trade structure with TSM is preferred. As a standalone short, asymmetry is too thin.

§ 02

Competitive position · every segment one-way negative.

5-axis · low across the board

5-axis moat scoring · INTC · per competitor.md Phase 3

Efficient scale is 1.

Foundry is not a natural monopoly geometry for Intel — TSMC already occupies that position. Trying to build a parallel scale curve with smaller volume is the structurally hardest game in semis.

Cost advantage is 2.

Lost the process-node cost-curve advantage to TSMC at 10nm/Intel 7. Cost-per-transistor flat-to-rising at N2/A16 erodes the historical "tick-tock" lever. Foundry losses persist ($2.3B operating loss Q3 2025).

Switching costs at 3 — but trajectory is one-way.

Datacenter x86 stickiness is real (validation cycles), which is why Intel still holds ~71% server unit share despite years of share loss. But the trajectory is one-way: each new instance class hyperscalers launch on ARM is a permanent migration, not a temporary blip.

Verdict Narrow · eroding. Five-year trend: every segment Intel competes in has lost share except where the floor is already at zero. The Q1 2026 earnings beat ($13.58B, EPS $0.29 vs $0.01 consensus) is a cyclical improvement — agentic-CPU demand pulling Xeon — not a structural share recovery.

§ 03

Supply chain · the strategic chokepoint is dependence on TSMC.

first commercial High-NA EUV customer

Counterintuitively, Intel's input position is better than the standard short framing implies. Intel is the first commercial High-NA EUV customer (EXE:5200B installed) at a moment when TSMC has deferred High-NA for A14. The competitive moat is eroding at the output layer (share, margin, ASP). The input layer is unremarkable to favorable. The short is a pass-through-power short and a capital-allocation short — not a supply-fragility short.

Diagram · supplier flow · risk-codedstrategic chokepoint = own dependence on TSMC

flowchart LR
  classDef sev fill:oklch(32% 0.10 28),stroke:oklch(60% 0.14 28),color:oklch(95% 0.05 28),font-family:'Geist Mono',font-size:11px
  classDef elev fill:oklch(35% 0.10 95),stroke:oklch(72% 0.13 95),color:oklch(95% 0.04 95),font-family:'Geist Mono',font-size:11px
  classDef ok fill:oklch(28% 0.06 230),stroke:oklch(58% 0.08 230),color:oklch(94% 0.04 230),font-family:'Geist Mono',font-size:11px
  classDef anchor fill:oklch(35% 0.14 28),stroke:oklch(72% 0.14 28),color:oklch(98% 0.05 28),font-family:'Geist Mono',font-weight:600,font-size:13px

  ASML[ASML EUV / High-NA
Intel = FIRST customer]:::elev
  KLA[KLA inspection]:::ok
  TEL[Tokyo Electron
coater-developer]:::ok
  AMAT[AMAT · LAM
standard]:::ok
  AJI[Ajinomoto ABF
industry-wide bottleneck]:::elev
  BESI[BESI · Shibaura
hybrid bonding]:::elev
  KIRYAT[Kiryat Gat Israel
human capital · reservists]:::elev
  TSMC_OUT[TSMC outsourced tiles
Lunar/Arrow/Panther Lake
$5.6B 2024 → $9.7B 2025]:::sev

  INTC{{Intel
Arizona · Oregon · Israel · Ireland
Foundry $50M Q2'25 ext.
$17.83B IFS internal}}:::anchor

  CCG[Client · PC OEMs
~46% rev]:::elev
  DCAI[Datacenter · AI
~24% rev]:::elev
  MBLY[Mobileye + Other
~5%]:::sev

  DELL[Dell ~17–19%]:::ok
  LEN[Lenovo]:::ok
  HP[HP · slipping]:::elev
  AWS[AWS Graviton · Cobalt]:::sev
  MS[MS Azure · Cobalt]:::sev
  GCP[Google · Axion]:::sev
  META[Meta · Vera CPU]:::sev

  ASML --> INTC
  KLA --> INTC
  TEL --> INTC
  AMAT --> INTC
  AJI --> INTC
  BESI --> INTC
  KIRYAT --> INTC
  TSMC_OUT --> INTC

  INTC --> CCG
  INTC --> DCAI
  INTC --> MBLY

  CCG --> DELL
  CCG --> LEN
  CCG --> HP
  DCAI -.-> AWS
  DCAI -.-> MS
  DCAI -.-> GCP
  DCAI -.-> META

Pass-through pricing power

5 / 5

5 = worst. The killer feature of the input-side analysis. GM 55.5% (2021) → 32.7% (2024) → 34.8% (2025) — a 2,000+ bp drawdown is the financial signature of pricing-power loss. Q2'26 GAAP guide steps down from Q1's 39.4% to 37.5%.

Foundry external

$50M Q2'25

Against $17.83B IFS segment (~99.7% internal). CFO Zinsner: external commitment "not significant right now." Falcon Shores cancelled, Gaudi 3 missed even the $500M target. Press-release-grade.

14A timeline

2028 · 2029

Risk production 2028, HVM 2029. Intel's last three node transitions all slipped multi-year. Binary outcome may not even resolve cleanly within a 12–18 month short horizon.

§ 04

Customer & end-market.

PC OEM concentration into a declining market

Top-3 OEM ~40% concentration into a structurally declining PC market.

CUSTOMER CONCENTRATION · per FY23 10-K

Dell

~19%

Top 3 (D+L+H)

~40%

Top 10 (est.)

~58%

Mobileye top-3

~53%

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. NVDA datacenter is running ~$30B/quarter. The merchant AI customer base never formed.

Hyperscalers have already qualified every alternative — AWS Graviton, Google Axion, Microsoft Cobalt, Meta Vera. The customer-relationship moat is structurally broken on the segment that contributes ~24% of revenue.

§ 05

Catalyst calendar · the regulatory cushioning question.

14A external commit + federal equity

Federal equity decision is the single largest "torpedo" risk for the short. The 14A external customer commitment is the resolution path.

§ 06

Kill criteria.

5 · observable · time-bounded

A second brand-name fabless customer commits to 14A by Q1 2027. Apple beyond entry-tier · NVIDIA wafer volume · AMD · Qualcomm · MediaTek · Broadcom. One Microsoft chip + DoD RAMP-C is press-release-grade; two is regime change.
Federal equity injection or CHIPS covenant loosening announced before Q3 2026 at terms that materially recapitalize (>$10B injection, multi-year covenant relief). Single regulatory event torpedoes the financing-stress timeline.
Intel beats AMD's datacenter growth in three consecutive quarters. Q1'26 was the first such beat in 19 quarters. One = cyclical. Three = structural inflection that invalidates the customer-erosion leg.
18A external foundry revenue exceeds $1B/quarter run-rate by Q4 2026. Current run-rate ~$222M/quarter. 4× scaling converts "press-release demand" to "commitment demand."
Gross margin expands above 42% GAAP for two consecutive quarters before end-2026. The line above which the structural pricing-power-loss thesis breaks.

"This is what compresses the PM conviction from 4 (the average of the analysts) to 3 (the timing-honest read)."— PM thesis · INTC · disagreement § 4