TSMC says its 4nm chip process is on track for risk production at end of 2021 and mass production in 2022, 3nm is expected in H2 2022, and 2nm is in development
Context & Ripple Effects
This update slots another rung into a ladder TSMC has been building in public for a year: it began installing 3nm fabrication lines back in mid-2020 [[a:954522]], and said its 5nm N5 chips would ship in products that year using 30% less power than N7 [[a:957148]]. Announcing 4nm as a derivative node between them keeps every customer transition incremental rather than a full-node jump.
The timing matters because TSMC was simultaneously expanding physically — construction started on its 5nm Arizona factory [[a:966962]], and final approval came through for a dedicated 2nm plant in Taiwan [[a:968977]]. Each roadmap statement here is a commitment that later coverage checks against.
First-order effects
- TSMC's customers get an interim 4nm option entering risk production at end of 2021 — a lower-risk migration path than waiting directly for 3nm mass production in 2022.
- The H2 2022 target for 3nm volume output, first set when the lines went in during 2020 [[a:954522]], is publicly reconfirmed, letting lead customers anchor their next product cycles to it.
Second-order effects
- A two-step year (N4 then N3) tightens design-win competition for TSMC capacity, and the later roadmap detail showing five planned N3-class iterations [[a:979998]] shows how TSMC monetizes each derivative rather than one big leap.
- The 2nm development claim forces capacity planning years out: equipment installs at the approved 2nm Taiwan plant were slated for 2023 [[a:968977]], so toolmakers and substrate suppliers align investment to TSMC's stated cadence.
Third-order effects
- The pattern — derivative nodes announced early, backed by pre-built fabs — concentrates leading-edge supply further around whoever executes the cadence, which is why TSMC could extend the same playbook to A16 and A14 timelines years later [[a:884977]].
- If the roadmap discipline holds while new fabs come online in Arizona and Taiwan, the structural outcome is leading-edge capacity becoming a contracted, multi-year commitment business rather than spot-buyable supply — the essence of the semiconductor capacity lag.
The trend: Leading-edge chipmaking is consolidating around TSMC's publicly promised node cadence, with derivative nodes and pre-committed fabs turning process leadership into multi-year customer contracts.