TSMC January revenue rises 37% on AI chip demand

TSMC reported a 37% rise in January revenue, reaching NT$401.3 billion (US$12.7 billion), driven by ongoing demand for AI-related chips.

The growth exceeded the company’s full-year forecast of 30%.

The January figure may be influenced by the Lunar New Year holidays, which fell in January 2025.

TSMC, a key supplier for Nvidia and Apple, is planning up to US$56 billion in capital expenditure for 2026, a 25% rise from 2025, primarily to support data center chip demand.

Despite the strong sales, concerns remain about the sustainability of AI-driven growth amid industry volatility and the potential for a tech cycle downturn.

The increased spending by major tech firms like Amazon and Meta has raised questions among investors about the long-term impact of AI investments.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

Growth clusters around high-end chip processes

Revenue gains are uneven. Most of the lift comes from the priciest chips used in AI-focused data centers.
In the fourth quarter of 2025, chips made on 7-nanometer processes or smaller made up 77% of TSMC’s total wafer revenue 1. The 3-nanometer process brought in 28% 1.
High-Performance Computing (HPC) drives most of this pull. The sector includes AI data center accelerators (specialized chips used to speed up AI training and inference in data centers).
In 2025, the HPC platform rose 48% year over year and delivered 58% of total revenue, while the smartphone platform increased 11% 1.

AI budgets are bending big tech cash flow and competition

TSMC’s expansion tracks a sharp jump in capital spending by top customers. Cloud and social media firms are building AI-capable data centers.
Amazon, Alphabet, Meta, and Microsoft are forecast to spend about $650 billion on infrastructure in 2026 2.
Analysts expect near-term strain. They project negative free cash flow at Amazon, and a nearly 90% drop at Alphabet 3.
Analysts expect these builds to widen an infrastructure competitive moat 3. TSMC’s CEO called AI customers “very rich,” said silicon remains the bottleneck, and said TSMC tracks power supply plus other infrastructure needs 1.