TSMC (台積電) eyes 30% plus revenue jump in 2026, pours money into 3nm capacity as AI demand surges

Taiwan Semiconductor Manufacturing Co (TSMC, 台積電) told investors it expects revenue to climb by more than 30% in 2026 as the world’s largest contract chipmaker rushes to expand capacity to meet booming artificial‑intelligence and high‑performance computing (HPC) demand. First‑quarter revenue was US$35.9 billion, up 40.6% year‑on‑year and slightly above the company’s guidance, underscoring how AI workloads are overtaking smartphone chips as the company’s primary growth driver.

Ramping for AI: bigger bets on 3‑nm and packaging

To feed hyperscale cloud and AI customers, TSMC has shifted capital intensity decisively upward. The company said it expects 2026 capital expenditure to reach the high end of its record US$52–56 billion range, and CEO C.C. Wei warned that capex over the next three years will be “significantly higher than the past three years.” It has been reported that TSMC is accelerating equipment deliveries and prioritizing 3‑nm node production and advanced packaging — investments designed to squeeze more performance out of each wafer. How fast can fabs be expanded? Fast enough to keep customers from hoarding capacity, apparently.

Tight supply, fierce competition and geopolitical headwinds

Wafer supply remains tight and advanced packaging capacity is particularly constrained, reportedly driving intense customer competition for slots. Management downplayed concerns about memory price swings and Middle East tensions but the broader backdrop matters: US export controls on advanced semiconductor equipment and chips bound for China, and wider US‑China tech competition, are reshaping where and how capacity is built. TSMC’s capital decisions now have strategic as well as commercial weight — investors and governments alike are watching where the new fabs and equipment go.

TSMC’s outlook crystallises a larger industry shift: the foundry world is moving from mass‑market mobile SoCs to highly specialised, high‑margin chips for AI and HPC. For Western cloud giants, Chinese cloud providers and governments weighing supply‑chain resilience, the question is no longer whether AI will drive demand — it’s who will secure the scarce advanced capacity first.