TSMC is already on track to ramp 2nm production by the end of 2025, but the company’s sights are set even further ahead. Its next big leap is the 1.4nm node—internally referred to as A14—and the foundry giant is preparing to kick off early production steps in Taiwan without leaning on the newest, ultra-pricey High-NA EUV scanners.
Instead of adopting High-NA EUV for A14, TSMC plans to push its current EUV infrastructure further using advanced multi-patterning. This approach demands more process steps and tighter control, but it lets TSMC avoid the staggering cost and supply constraints of first-wave High-NA tools. The payoff could be substantial: A14 is expected to improve power efficiency by up to 30 percent, a boon for everything from flagship smartphones and laptops to AI accelerators and data center silicon.
Here’s how the roadmap takes shape, based on recent reporting and industry chatter:
– Groundbreaking for a new 1.4nm fab in Taichung is targeted for the end of this year.
– R&D for A14 is centered in Hsinchu, while hiring is underway in Taichung; construction permits for three buildings were issued in August.
– Mass production of 1.4nm is slated for the second half of 2028, aligning with TSMC’s longer-term cadence for next-gen nodes.
– The initial investment could reach roughly NT$1.5 trillion (about $49 billion), including plans to purchase around 30 EUV systems in 2027.
– TSMC is expected to bypass ASML’s High-NA EUV scanners—priced at about $400 million per unit—and rely on complex multi-patterning strategies with its existing EUV equipment.
Why skip High-NA EUV now? While High-NA promises improved resolution and simplified patterning at leading-edge nodes, the first generation of these machines is extraordinarily expensive and still early in its lifecycle. TSMC appears confident it can deliver A14 at scale using today’s EUV tools, reinforced by deep experience in multi-patterning. The trade-off is complexity: more masks, more process steps, and potentially lower initial yields. Expect a ramp characterized by trial-and-error refinements as engineers tune the process. The crucial advantage for TSMC is that it already operates a large, mature EUV fleet and has years before 1.4nm enters volume production.
For context, multi-patterning has also been used elsewhere to stretch existing lithography to smaller geometries, but TSMC’s extensive EUV know-how and supply chain leverage could help mitigate the cost and yield penalties over time. With 2nm coming online first, the foundry has a meaningful buffer to de-risk A14 before the planned 2028 mass-production window.
What to watch next:
– The 2nm ramp in late 2025 as a precursor to A14.
– Construction progress in Taichung and R&D milestones in Hsinchu.
– 2027 equipment orders, a key indicator of the final A14 toolset.
– Early power and performance disclosures that validate the promised efficiency gains.
Bottom line: TSMC is doubling down on a pragmatic path to 1.4nm, trading the headline-grabbing allure of High-NA EUV for a more measured, cost-aware strategy built on multi-patterning and an established EUV base. If the company executes, A14 could land with strong efficiency improvements and keep TSMC in the driver’s seat at the cutting edge of semiconductor manufacturing.
