A capacity-constrained AI supply chain supports multi-year revenue compounding, while heavy capex and geopolitics limit how far the valuation can expand. Upside comes from converting scarcity into contracts and assurance services that reduce cyclicality.

TSMC remains the AI era’s “compute refinery”: if leading-edge + advanced-packaging scarcity persists and TSMC productizes trust (outcome-based capacity, verified packaging, sovereign supply), it can compound revenue into 2031 while sustaining a premium multiple despite heavy capex and geopolitics.

Compute is the binding input of the Last Economy; TSMC owns the hardest-to-scale conversion of capex+know-how into reliable leading-edge output, with high switching costs and trust-based network capital across the ecosystem.

TSMC’s 2031 upside is less “new product” and more scarcity-powered throughput + mix: leading-edge logic and advanced packaging stay structurally constrained by tool supply, yield ramps, and qualification cycles, supporting durable utilization and pricing. The incremental non-linear lever is moving up the stack from “wafers sold” to “outcomes guaranteed” (take-or-pay capacity blocks, delivery SLAs, and security/assurance attach), which should reduce cyclicality and defend margins in a compute-hungry world even as overseas capacity expands.

The swing factors are (1) utilization under a rising depreciation load, (2) N2 and advanced-packaging ramp execution versus demand, and (3) geopolitical/policy permissioning that can both raise costs (overseas dispersion) and impose a persistent risk discount.

Modestly stronger overseas leading-edge optionality (Japan 3nm) supports sovereign-capacity upside; core scarcity-throughput thesis unchanged.

Terminal multiple assumption kept disciplined given policy/geopolitical discount and sustained high depreciation burden despite strong demand signals.

$419.81