TSMC: moat, 2nm and new pressures

TSMC did not just have a good 2025. It widened the gap. The company’s 2025 revenue rose 31.6% to NT$3.81 trillion, and market trackers say it ended the year with roughly 70% of the global foundry market while Samsung slipped into the low single digits of share at the leading edge (tsmc.com) (counterpointresearch.com) (trendforce.com). That is the moat in plain terms. TSMC is not merely bigger than its rivals. It is compounding faster than they are, which matters because the next jump in chipmaking is getting harder, not easier. That is why 2nm matters so much. TSMC says its N2 process entered volume production in the fourth quarter of 2025, right on schedule, and it is the company’s first major shift to nanosheet transistors after years of FinFET scaling (tsmc.com). This is not a branding tweak. It is a transistor redesign aimed at squeezing out more performance and lower power as smartphones, AI accelerators and high-performance processors all hit the same wall: electricity and heat. TSMC’s own technical disclosures describe N2 as a full-node advance, with better density and energy efficiency than the 3nm generation it follows (tsmc.com) (tomshardware.com). The timing is not accidental. Foundry growth in 2025 was driven less by ordinary handset cycles than by AI. Counterpoint says TSMC held 72% of the pure-play foundry market by late 2025, helped by strong 3nm demand in phones and PCs, heavy 4nm and 5nm demand from Nvidia’s Blackwell GPUs, and continued contribution from CoWoS advanced packaging (counterpointresearch.com). TSMC had already told investors in April 2025 that it was working to double CoWoS capacity that year to keep up with customer demand (investor.tsmc.com). That helps explain the strange shape of the business now. The most important chips are no longer just about the node. They are about the node, the packaging, the yield curve, and the ability to deliver all of it at once. That same complexity is why TSMC’s problems are changing. The old fear was that China might threaten Taiwan with ships and missiles. That fear is still there. But the newer pressure is more granular and in some ways more practical: talent poaching, cyberintrusions, and attempts to extract process knowledge without invading anything. Reuters reported on April 6 that Taiwan’s National Security Bureau warned China is trying to obtain Taiwan’s advanced chip technology and talent through indirect channels, including poaching, technology theft and controlled-goods procurement (usnews.com). The same report said Taiwan’s government service network faced more than 170 million intrusion attempts in the first quarter of 2026 (usnews.com). That warning landed on top of reports that TSMC had already dealt with an attempted theft of 2nm-related trade secrets inside Taiwan. Coverage in 2025 said employees were fired and legal action followed after the company discovered efforts to remove sensitive information tied to its most advanced process technology (macobserver.com) (channelnewsasia.com). In other words, the crown jewels are no longer just fab buildings. They are recipes, yield data, process integration tricks, and the people who know where the hard parts are. That helps explain why TSMC’s expansion abroad has become part industrial policy and part insurance policy. In March 2025, TSMC announced plans to expand its U.S. investment to $165 billion, adding three more fabs, two advanced packaging facilities and an R&D center in Arizona on top of projects already underway (tsmc.com). But even that does not erase Taiwan’s central role. TSMC’s own 2nm page says Fab 20 and Fab 22 are the production facilities for N2, and N2P is scheduled for volume production in the second half of 2026 (tsmc.com). The next stage of the story is still anchored on those fabs, where the real contest is not just who can announce 2nm first, but who can protect it long enough to ship it.