Compute & Infrastructure
Designs and manufactures CPUs for PCs and servers and is building a contract foundry; a long-time leader now racing to catch up in AI.
{'verdict': '3 signals sit in the elevated band: the AI-monetization gap, operating leverage and the valuation premium vs fundamentals. This does not trip the convergence flag. Ranks 15th of 68 on composite fragility (F\xa056.45), below MU and BigBear.ai.', 'as_of': '2026-07-11', 'source': 'engine-restatement (T1)', 'snapshot': {'composite_f': 56.45, 'n_elevated': 3, 'convergence': 'moderate', 'rank': 15, 'elevated': ['the AI-monetization gap', 'operating leverage', 'the valuation premium vs fundamentals']}}
Intel Foundry lost $10.318B in FY2025 on top of $13.291B in FY2024. Gross capex in FY2025 was $14.6B. Total debt is ~$46.6B. How does Intel service ~$46.6B in debt while burning $10B+ per year at the foundry level — and what is the external foundry revenue that is supposed to justify this?
The numbers are confirmed primary: Foundry operating loss FY2025 -$10.318B, FY2024 -$13.291B (Intel FY2025 and FY2024 earnings releases). Gross capex FY2025 $14.6B (net ~$11.2B), down from $23.944B FY2024. Total debt Dec 2025 ~$46.6B ($2.496B short-term + ~$44.1B long-term). External Intel Foundry third-party revenue is NOT SOURCED — Intel does not disclose it separately; reported foundry revenue is overwhelmingly intersegment. The CHIPS Act $8.9B equity injection (Aug 2025 at $20.47/share, 9.9% stake — secondary) and SCIP partner contributions ($11.861B FY2024 cash flow from Brookfield and Apollo — primary) partially de-risk the balance sheet, but the foundry loss structure remains the central fragility per this sheet.
Intel guided approximately $500M in Gaudi 3 AI sales for H2 2024, then discontinued the Gaudi roadmap post-Gaudi 3. External foundry design-win revenue commitments are NOT SOURCED. What is the actual evidence of external AI demand for Intel silicon?
The Gaudi situation is confirmed: $500M H2 2024 guidance came from the Q2 2024 earnings call. Gaudi revenue is NOT broken out within DCAI ($16.919B FY2025 total includes Xeon CPUs, Gaudi, Altera, and other DC products). Gaudi roadmap was discontinued per the sheet (secondary: More Than Moore). The positive FY2025 story is DCAI +32% to $16.919B — driven by Xeon server CPU refresh, not dedicated GPU accelerators. External foundry customer named wins and revenue commitments are NOT SOURCED. CEO Tan's equity accumulation (4.2M options Mar 2025, 774K Feb 2026, zero code-S sales as CEO) is the contra-insider signal, but it does not source external foundry traction.
Intel's January 2023 life extension from 5 to 8 years on fab machinery reduced 2023 depreciation by ~$4.2B including ~$2.5B of gross-margin benefit — per Intel's own 8-K. Then Q3 2024 produced $2.3B in impairments plus $992M in accelerated depreciation when Intel 7 demand fell short. Is that extension-then-impairment sequence a depreciation-integrity issue or just textbook accounting?
Both facts are primary: Intel's own 8-K quantified the 5→8-year extension benefit ($4.2B reduced depreciation, $2.5B gross margin, ~$0.4B R&D, ~$1.3B inventory) effective Q1 2023. The Q3 2024 impairment ($2.3B non-cash + $992M accelerated depreciation on Intel 7 assets) is confirmed from the FY2024 10-K notes, recorded when Intel 7 capacity exceeded projected demand. The sheet frames this as "classic depreciation-then-reality sequence": the extension flattered FY2023 earnings during peak foundry buildout, and the 2024 impairment was a partial forced unwind. With $50.4B in construction-in-progress not yet depreciating (Dec 2024 balance sheet, primary), the next wave of this dynamic has not yet hit the income statement.