TSMC has chosen to extend its reliance on Low-NA EUV lithography for at least one more node, delaying the transition to High-NA systems that could cost up to €350 million per machine. The move is part of a broader strategy to optimize capital expenditure while maintaining aggressive scaling goals.

The decision to stick with Low-NA EUV—paired with multi-patterning techniques—allows TSMC to achieve performance gains similar to High-NA without the financial burden. However, the approach comes with tradeoffs: while effective for current nodes like A13, it may limit flexibility as process requirements tighten at 1 nm and below.

Why Low-NA Still Works

The core of TSMC's strategy is leveraging existing Low-NA EUV tools more efficiently. By using multi-patterning—where the machine makes two passes to etch a single layer—the company can replicate some benefits of High-NA exposure without the cost. This mirrors Intel's earlier approach with its 10 nm node, though Intel later pivoted to High-NA for its 14A process.

Yet, TSMC is not alone in this path. Other foundries are also exploring ways to extend Low-NA EUV, but the industry's shift toward High-NA—driven by demand for finer patterning at advanced nodes—means this strategy won't last forever. The question now is how long TSMC can sustain this advantage before being forced to adopt High-NA, a move that could reshape its roadmap and market positioning.

TSMC Extends Low-NA EUV Lead, Pushing Back High-NA Transition

Key Specs

  • Low-NA EUV: Multi-patterning (two passes per layer) to match High-NA performance.
  • High-NA EUV Cost: €350 million (~$410 million) per machine, significantly higher than Low-NA alternatives.
  • Node Focus: Low-NA effective for A13 and similar nodes; High-NA needed for 1 nm and below.

The reality check is that while TSMC's approach works today, the industry is moving toward High-NA. The risk is clear: if TSMC delays too long, it may find itself in a position where catching up becomes more expensive and complex than leading from the start.

For now, TSMC remains confident in its ability to extract maximum performance from Low-NA tools. But as competitors like Intel and Samsung ramp up High-NA adoption, the window for this strategy could close faster than expected. The balance between cost and innovation will define whether this gamble pays off—or leaves TSMC playing catch-up.