🇨🇳TechNode•Freshcollected in 56m
TSMC skips €350M ASML High-NA tool

💡TSMC rejects €350M tool, signaling sub-1.4nm delays for AI chips.
⚡ 30-Second TL;DR
What Changed
TSMC has no plans to buy ASML High-NA EUV machines.
Why It Matters
This could delay TSMC's advanced node transitions, affecting supply of high-end AI GPUs from Nvidia and others dependent on sub-2nm processes. Competitors like Samsung or Intel may gain edge in lithography adoption.
What To Do Next
Compare Samsung's High-NA EUV roadmap for hedging AI chip production risks.
Who should care:Enterprise & Security Teams
🧠 Deep Insight
AI-generated analysis for this event.
🔑 Enhanced Key Takeaways
- •TSMC is prioritizing the optimization of its existing 'Low-NA' EUV fleet, specifically utilizing multi-patterning techniques to achieve sub-1.4nm nodes without the immediate capital expenditure of High-NA systems.
- •Industry analysts suggest TSMC's decision is driven by a 'cost-per-wafer' calculation, where the throughput and yield maturity of current EUV machines currently offer a more favorable ROI compared to the early-stage High-NA infrastructure.
- •Intel remains the primary early adopter of ASML's High-NA EUV (EXE:5000/5200 series), positioning its foundry strategy around the early integration of these tools to gain a potential lithography resolution advantage.
📊 Competitor Analysis▸ Show
| Feature | ASML High-NA EUV (EXE:5000) | ASML Low-NA EUV (NXE:3800E) |
|---|---|---|
| Numerical Aperture | 0.55 | 0.33 |
| Resolution | ~8nm | ~13nm |
| Target Nodes | < 2nm (A14/A10) | 3nm / 2nm |
| Estimated Price | > €350 Million | ~€150-200 Million |
| Primary Adopter | Intel | TSMC, Samsung, Intel |
🛠️ Technical Deep Dive
- High-NA (Numerical Aperture) EUV utilizes a 0.55 NA lens compared to the 0.33 NA of standard EUV, allowing for higher resolution patterning without the need for complex multi-patterning steps.
- The EXE:5000 series requires a significant redesign of the wafer stage and reticle handling systems due to the anamorphic lens design, which shrinks the field size by half in one dimension.
- High-NA systems require higher power consumption and specialized photoresist materials to compensate for the reduced depth of focus inherent in the higher NA optics.
🔮 Future ImplicationsAI analysis grounded in cited sources
TSMC will face increased process complexity at the 1nm node.
By relying on multi-patterning with Low-NA tools instead of High-NA, TSMC must manage more lithography steps, which increases the risk of overlay errors and reduces overall wafer throughput.
Intel's foundry margins will be under pressure compared to TSMC.
The massive capital depreciation costs associated with early High-NA adoption will weigh heavily on Intel's foundry unit costs compared to TSMC's more mature, lower-cost equipment base.
⏳ Timeline
2023-12
ASML ships the first High-NA EUV system (EXE:5000) to Intel's Oregon facility.
2024-05
TSMC confirms it has successfully produced 2nm test chips using existing Low-NA EUV technology.
2025-09
TSMC begins volume production of its N2 (2nm) process node.
2026-04
TSMC officially announces at the North America Technology Symposium that it will skip High-NA EUV for the immediate future.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: TechNode ↗

