TSMC posts record quarterly profit on $46.7 billion revenue
TSMC, the world's largest contract chipmaker, grew third-quarter revenue 41% to $46.7 billion, beating its own guidance. Gross margin dipped slightly as 2-nanometre production ramped up.
What happened
TSMC reported third-quarter revenue of NT$1,494 billion and net income of NT$763 billion on 15 October, a record quarterly profit. In US dollars revenue reached $46.72 billion, up 41.2% year on year and 16.2% from the previous quarter. Earnings per ADR were $4.60.
The numbers
TSMC had guided to $44.6-45.8 billion, so it beat the top of the range by about $1 billion. Gross margin was 66.4%, operating margin 58.7% and net margin 51.1%. Fourth-quarter revenue guidance is $52.3-54.3 billion with a 65-67% gross margin.
Background
Gross margin fell 1.3 points from the previous quarter, which the CFO attributed to the 2-nanometre ramp. New process nodes start with higher scrap rates and heavy depreciation, so they dilute margins at first and lift them as they mature.
Why it matters
TSMC makes the most advanced chips for Nvidia, Apple, Google and Amazon, so its results are a direct read on AI demand. Demand is clearly growing, yet the stock still fell during the week's $1 trillion chip selloff as investors focused on how the build-out will be financed.
Who buys what
A growing share of TSMC's revenue comes from high-performance computing, meaning data-centre chips. Nvidia GPUs, Google TPUs, Amazon Trainium and Microsoft Maia all come out of the same fabs. That makes TSMC a neutral supplier in the AI race: whoever wins, TSMC most likely makes the chip. Smartphone processors are also moving to 3 and 2 nanometres. Its main weakness is geography. Most output is still in Taiwan, which keeps geopolitical risk on the table. The Arizona fabs were built to reduce that risk, but their capacity is still small compared with Taiwan.
What comes next
Guidance implies another 12-16% revenue growth next quarter. Margin pressure from 2 nm is expected to last a few quarters.
What it means for your business
Money is still pouring into AI infrastructure, so the cost of running models will keep falling. Review your AI tool pricing every six months, and watch on-device AI, which lets sensitive work stay off the cloud.
