TSMC Q2 2026 Earnings: Net Profit Up 77% to a Record as AI Demand Lifts Margins to 67.7%, 2nm Revenue Debuts

TSMC, the world’s largest contract chipmaker, reported its second-quarter 2026 results on July 16¹. Revenue came in at NT$1,270.38 billion ($40.20 billion, up 33.7% year over year in US dollar terms), and net income reached NT$706.56 billion, up 77.4% from a year earlier¹. Both revenue and profit are reported to be records³.

TSMC manufactures chips for NVIDIA, Apple and most of the AI industry⁴, making its earnings the single most-watched data point for judging whether the massive investment in generative AI is still translating into real semiconductor demand.

Inside the Numbers: 67.7% Gross Margin, Above the Company’s Own Guidance

Gross margin for the quarter was 67.7%, operating margin 60.3% and net profit margin 55.6%¹. Gross margin exceeded the top of the company’s own guidance range of 65.5–67.5%³. Growth accelerated sequentially too — revenue rose 12.0% and net income 23.4% from the first quarter¹ — landing at the top of the $39.0–40.2 billion revenue range TSMC projected in April⁴.

The engine is high-performance computing (HPC), the segment that includes AI chips: it reportedly accounted for 66% of total revenue, up 20% from the previous quarter³. Smartphones fell to 22%, down 4% sequentially³ — the shift from “the iPhone chipmaker” to “the AI infrastructure company” is now visible in the mix.

By process node, the leading-edge 2-nanometer generation appeared in the results for the first time, at 3% of total wafer revenue¹. The 3nm node contributed 30%, 5nm 33% and 7nm 11%, with advanced technologies (7nm and beyond) making up 77% of wafer revenue¹. Finer processes yield faster, more efficient chips, and that concentration in leading-edge nodes is the source of TSMC’s pricing power.

Q3 Guidance: Up to $45.8 Billion on a “Steep Ramp-Up” of 2nm

“Our business in the second quarter was supported by strong demand for our leading-edge process technologies,” said Wendell Huang, Senior VP and Chief Financial Officer. “Moving into third quarter 2026, we expect our business to be supported by continued strong demand for our leading-edge process technologies, including the steep ramp-up of our 2-nanometer technology.”¹

Third-quarter guidance calls for revenue of $44.6–45.8 billion (assuming 32 NT dollars per US dollar, with gross margin of 65–67%)¹ — implying another double-digit sequential increase. The 2nm generation is expected to power Apple’s next chips and new AI accelerators, and TSMC describing its ramp as “steep” is itself a signal about the strength of leading-edge demand.

The Backstory: The Supply Chain Keeps Confirming AI Demand Is Real

Since the generative AI boom began, the recurring question has been whether AI investment is a bubble. The evidence to test it comes from actual results at each layer of the supply chain. In memory, SK Hynix rode AI-driven HBM demand to the largest foreign IPO in US history. In design, NVIDIA has expanded from GPUs into full AI infrastructure. TSMC, which manufactures for nearly all of them, growing profit 77.4% shows the core of that chain still running at full capacity.

The demand side points the same way. This week, NVIDIA rolled out a physical AI ecosystem across Japan, adding large new chip orders such as RIKEN’s supercomputers and Mizuho’s AI factory. At the same time, pushback against the power and social costs of AI infrastructure is growing — New York just paused permits for large new data centers — so record supply-chain earnings and regulatory friction are now advancing in parallel.

Why It Matters: Primary Data for AI Investment Decisions

For readers making AI investment or adoption decisions, this report is primary evidence that AI demand is not slowing at the manufacturing layer. If guidance holds, tight chip supply and premium pricing for leading-edge nodes will likely persist, which means the underlying cost of AI services is unlikely to fall quickly.

One caution: TSMC’s results prove that AI companies keep spending on silicon — not that their AI spending is profitable. The record numbers upstream (chip manufacturing) and the pace of monetization downstream (AI services) should be evaluated separately. The next checkpoints are how far 2nm’s revenue share climbs in Q3, and whether TSMC’s biggest customers adjust their capital spending plans.