Taiwan Semiconductor Manufacturing Company reported first-quarter 2026 net profit of NT$361.6 billion ($11.2 billion), a stunning 58% year-over-year increase that demolished analyst estimates, as surging orders for the world's most advanced AI chips propelled revenue to a record NT$839.3 billion. The results cemented TSMC's position as the irreplaceable backbone of the global artificial intelligence buildout and sent shares climbing more than 4% in Taipei trading.
CEO C.C. Wei said demand for TSMC's 3-nanometer and next-generation 2-nanometer process nodes — used to manufacture chips for Nvidia's Blackwell GPU family, Apple's A-series processors, and AMD's latest data center accelerators — remains constrained only by the company's production capacity. TSMC plans to invest $38 billion in capital expenditure in 2026, with the bulk directed toward expanding advanced node capacity in Taiwan and its new Arizona fabs.
AI Orders Dominate Revenue Mix
High-performance computing — the segment encompassing AI accelerators and data center chips — accounted for 59% of TSMC's total wafer revenue in Q1, up from 46% in the same quarter a year ago. Smartphone chips, historically TSMC's largest revenue category, fell to a 28% share. The shift reflects the breathtaking pace at which hyperscalers including Microsoft, Amazon, Google, and Meta are pouring capital into AI infrastructure.
Gross margin expanded to 58.8% from 53.1% a year ago, beating the guidance midpoint of 57.5%. The improvement was driven by higher utilization of leading-edge nodes, favorable pricing on new contracts, and a weaker New Taiwan dollar that boosted the U.S. dollar value of domestic revenues.
Guidance Points to Continued Acceleration
For the second quarter of 2026, TSMC guided revenue of $28.4 billion to $29.2 billion — at the midpoint, a 38% increase over Q2 2025 — and gross margin of 57% to 59%. The outlook exceeded the Wall Street consensus by approximately $1.5 billion at the midpoint, triggering a broad rally in semiconductor stocks including Nvidia, ASML