Net Profit Surges 77%
Q2 revenue was NT$1.27 trillion ($40.2 billion), up about 34% year over year; net profit was NT$706.6 billion, up 77% year over year, substantially exceeding LSEG SmartEstimate's NT$632.6 billion expectation; gross margin was 67.7%. By process, 3nm was 30% of wafer revenue and 5nm 33%, with the just-ramping 2nm already 3%. High-performance computing (HPC, the segment where AI chips sit) was 66% of total revenue, the absolute mainstay of growth.
Guidance More Aggressive Than the Results
Q3 revenue guidance is $44.6–45.8 billion, up over 10% sequentially; CFO Wendell Huang said Q3 will benefit from "the steep ramp of 2nm technology." More important is the full-year view: the USD revenue growth target was raised from "over 30%" to "over 40%," with capital expenditure raised substantially in tandem. The company had previously disclosed that N3 capacity and CoWoS advanced packaging are booked full through year-end.
Why This Report Is a Hard Metric
TSMC fabricates nearly all leading AI chips—NVIDIA, Google TPU, and everyone's in-house inference chips all ship from here—so its actual shipments and guidance are the hardest data on "whether AI compute demand has peaked." This report echoes ASML raising its guidance the day before: from equipment to foundry, the upstream chain is all continuing to accelerate expansion to meet demand. The direct implication for the industry is that the compute supply strain won't ease in the short term, and model vendors' inference costs and queuing problems will persist for a while.
via: TSMC official earnings and investor call (2026-07-16), Reuters / Benzinga reporting