- 2026 is the commercial inflection point for Intel Foundry as 18A ramps with improving yields but profitability and external customer volume remain unproven.
- Intel will advance 14A only with firm anchor-customer demand, otherwise extending 18A variants to limit capex risk.
- CHIPS Act and U.S. strategic backing help scale domestic capacity, but customers will still decide based on cost, yield, and reliability.
- TSMC’s lead and entrenched demand make Intel’s bid to become a credible second-place foundry plausible but uncertain.
Read More
Intel’s foundry model has moved beyond proving its technical capabilities (fab construction, node development) and is entering its crucial commercial test phase in 2026. The company has made tangible progress with its 18A node, which incorporates RibbonFET and PowerVia technologies, started high-volume manufacturing in late 2025, and is reportedly seeing yield improvements of roughly 7% per month, likely putting 18A yields in the 65–75% range entering 2026. This level is approaching what is needed for cost efficiency and competitive pricing. [Primary analysis],
However, these technical advances have not yet translated into strong external customer commitments. While Intel reports over $15 billion in expected lifetime deal value with external clients, most relationships are exploratory. Internally, Intel’s own demand—whether in PCs, internal products, or defense/governmental contracts—is dominant. Material external volume contracts with hyperscalers or major fabless chip designers have not yet materialized at scale. [Primary analysis]
The 14A process represents a strategic crossroad. Intel is using a more cautious investment posture: it intends to scale 14A only if anchor customers and high-margin external demand are secured. This denotes a more disciplined capital allocation model than past cycles. Absent strong demand, Intel may instead extend and refine its 18A-P (performance variant) through the late 2020s rather than press aggressively for 14A. [Primary analysis]
Financial discipline remains in focus. Intel’s Foundry division suffered operating losses (~US$7 billion in 2023) and continued to bleed through 2024–2025. Management projects that the worst of these losses has either happened or will happen in 2024, and that break‐even or better operating margins are a multi‐year goal—targeting ~40% non‐GAAP operating margin (30% for foundry) and 60%+ gross margins by 2030. [Secondary sources: Intel filings],
Geopolitical conditions inject both opportunity and constraint. U.S. government support—through a ≈10% passive ownership stake and CHIPS Act incentives—adds strategic weight to Intel’s foundry ambitions. The company is leveraged as a domestic “escape valve” should tensions with Taiwan or China hamper reliance on TSMC. But politics cannot substitute for market trust: foundry clients care about cost, yield, node leadership, and reliability. [Primary analysis],,
Competition remains both fierce and ahead. TSMC continues to lead in node maturity (3nm, 2nm paths), has long-standing anchor customers in AI, and benefits from scale. Samsung is still in advanced GAA/RibbonFET development. Against that backdrop, Intel is seen by analysts (e.g., KeyBanc, J.P. Morgan) as potentially moving into second place among foundries, but only if it can further accelerate yield gains, scale external demand, and control costs. [Secondary sources],
Strategic implications: Intel’s momentum creates a window during 2026 to validate its foundry model—success in securing anchor external customers (for 18A, and especially 14A), ramping yields, and managing capital intensity will define whether it becomes a scaled competitor to TSMC or remains behind. Potential strategic moves include pricing incentives to win design wins, leveraging government contracts, possibly forming joint ventures to spread risk (as with proposals from TSMC and chipmakers), and tailoring investment timing to demand.
Open questions: Will Intel land major design wins from Apple, Nvidia, or other hyperscalers for 18A or 14A? At what yield and cycle time will 18A become commercially competitive? Will the company’s capex and capacity commitments align with demand rhythms? And finally, will 2026 mark steady progress or another cycle of overreach?
Supporting Notes
- Intel began high-volume manufacturing (HVM) for the 18A node in late 2025 and is improving yields by ~7% per month; KeyBanc estimated yields at ~55% in mid-2025, suggesting entering 2026 with yields of 65–75% if trends hold.
- Intel’s Foundry business reported operating losses of approximately US$7 billion in 2023, with ongoing losses through 2024–2025; external customer revenue still significantly lags internal demand.
- Intel has projected a foundry backlog (contractual or expected) of over US$15 billion from external clients.
- The company has committed over US$100 billion since 2021 towards global manufacturing expansion, especially fabs in Arizona and Ohio.
- Intel’s financial targets: non-GAAP gross margin of ~60%, operating margin ~40% for its product segments by 2030; Foundry business is targeted to move toward break-even somewhere between present and 2030, with non-GAAP operating margin of ~30% at that time.
- Regarding 14A process: investment and capacity expansion depend on external customer commitments; Intel’s internal capacity isn’t yet sufficient for third-party 14A clients absent such demand.
- Market share and competitive pressure: TSMC continues to lead in advanced nodes; Intel is projected by some analysts to become the second-largest foundry if 18A progress continues. J.P. Morgan expects TSMC to maintain over 95% share in advanced 2nm chip manufacturing.,