Last price
$68.50
CPU & GPU enabler for AI; execution layer dependence is real.
Live quote sourced from Yahoo Finance. Prices cited in narrative below reflect the original memo date and may be stale.
Intel designs and manufactures CPUs (Xeon for data centers), GPUs (Arc), and custom ASICs. All autopilot platforms require semiconductor horsepower. However, Intel's core business is selling chips, not capturing the services-budget upside. The thesis lens: Intel is an enabler, not an operator.
Intel is in the hardest strategic position in semis — IDM 2.0 (foundry pivot) is a multi-year bet with execution risk, while the core merchant-CPU franchise is collapsing under Arm and custom-silicon pressure. The company simultaneously has picks-and-shovels upside (foundry for custom ASICs) and existential downside (CPU moat erasure). Read this as two bets on one ticker.
Intel's foundry business (Intel 4, Intel 3 nodes ramping 2026-2027) could become a credible alternative to TSMC for hyperscaler custom-silicon manufacturing. If hyperscalers diversify ASIC foundry (meta/Google/AWS split orders between TSMC and Intel 50/50), Intel's bare-wafer-fab capacity becomes valuable picks-and-shovels. Additionally, Gaudi AI accelerators are volume sampling; early indications show modest but real traction in hyperscaler labs.
| Segment | Revenue % | AI exposure | Verdict |
|---|---|---|---|
| Client CPU (PC) | ~30% | Declining | Headwind |
| Data-center CPU | ~35% | Eroding to AMD | Headwind |
| Foundry/Gaudi/other | ~35% | Upside optionality | Tailwind |
If hyperscalers diversify ASIC manufacturing away from TSMC concentration risk, Intel's Intel 4 node (2026-2027) becomes strategically valuable. Even 15-20% share of custom-ASIC foundry is $5-8B annual revenue at 35%+ gross margin.
Early indications show Gaudi competes favorably on inference cost and customization vs. NVIDIA for certain workloads. This is a niche play, but volume growth is possible.
Intel received $8.5B+ in subsidies through 2025. Additional funding is politically viable. This partially derisks the foundry pivot.
Intel is bleeding $2-3B annually in operating losses while betting $100B on foundry. If foundry delays further, the company runs out of time and capital.
TSMC is shipping N3 in volume; Intel is ramping Intel 4 in 2026-27 (equivalent to ~N5 maturity). Hyperscalers will use Intel for secondary nodes, not leading-edge — lower ASP and lower margin.
NVIDIA's CUDA ecosystem and software maturity are insurmountable. Gaudi is 3-4 years behind in software tooling. Hyperscalers will use Gaudi for <10% of inference, not 30%.
Trading at 8-10x forward earnings (depressed multiples), Intel has limited upside if foundry succeeds and catastrophic downside if it stalls. The binary nature makes this a speculation, not an investment.
Intel is the inverse of NVDA — maximum structural risk paired with maximum upside optionality. The Sequoia thesis implies foundry diversification away from TSMC (hyperscaler risk mitigation), which is a tailwind for Intel's IDM 2.0 pivot. However, the core CPU business is imploding, and foundry execution is unproven and delayed. This is not a thesis call; it is a restructuring speculation bet on whether Intel's foundry ramp can offset CPU losses. Verdict: Watch for thesis investors — restructuring optionality keeps it watchable, but the core case is a 5-year foundry-ramp bet, not a thesis call.
Intel's strength in CPUs and foundry aspirations matter for AI infrastructure, but services-budget capture remains with platform operators.