• TRM$ 3.461,92 +1,37 % +$ 46,82TRM oficial (Superintendencia Financiera), vigente del 21 nov al 23 nov
  • Euro$ 3.804,64 +1,37 % +$ 51,45Referencia: TRM cruzada con la tasa del Banco Central Europeo del 20 nov; no es la cotización de mercado
  • Peso mexicano$ 181,56 +1,37 % +$ 2,45Referencia: TRM cruzada con la tasa del Banco Central Europeo del 20 nov; no es la cotización de mercado
← Volver a la comunidad
Tecnología

Why is TSMC skipping ASML's $380M High-NA EUV machines at 1.4nm while Intel gambles its foundry turnaround on them?

TSMC is bypassing ASML's $380 million High-NA EUV lithography machines for its initial 1.4nm A14 node because 0.33-NA multi-patterning with curvilinear computational lithography delivers superior wafer unit economics today, whereas Intel is betting its 14A foundry turnaround on 0.55-NA single-exposure optics.

As the semiconductor industry enters the angstrom era, the capital expenditure divide between the two leading foundries centers on optical physics versus balance-sheet depreciation. ASML's Twinscan EXE:5000 and EXE:5200 systems upgrade the numerical aperture from 0.33 to 0.55, shrinking single-exposure resolution from 13nm down to 8nm and eliminating costly quadruple-patterning masks. However, achieving a 0.55 NA requires anamorphic lenses that halve the exposure reticle field size—forcing chipmakers to stitch dies together for massive multi-reticle AI GPUs—while adding nearly $400 million in tool depreciation per scanner to wafer manufacturing costs.

Beyond CapEx math, sub-2nm lithography is colliding with a fundamental quantum bottleneck: photon shot-noise and stochastic defects. Because High-NA spreads extreme ultraviolet photons across ultra-thin metal-oxide photoresists with shallow depth of focus, random statistical fluctuations in photon absorption cause micro-bridges, line-edge roughness, and missing contact vias that can tank die yields. By stretching existing 0.33-NA scanners with inverse lithography technology (ILT), TSMC protects its 55%+ gross margins while letting Intel absorb the early learning curve of High-NA resist chemistry. Will Intel's early $380M High-NA EUV bet capture AI foundry customers at 14A, or will TSMC's cost-optimized 0.33-NA roadmap win on wafer yield and price?

Which lithography strategy will win the 1.4nm (A14/14A) foundry race?

0 votos

Comentarios

Sé respetuoso. Los comentarios que reporten varias personas se ocultan hasta que los revise un moderador.

Para leer no necesitas cuenta; solo para comentar. Entras con un código que te llega al correo o con tu cuenta de Google. Mostramos solo el nombre que elijas; nunca tu correo.