TSMC posts 35% surge
TSMC reported a 35% revenue jump tied to AI demand and framed the result as the start of a 2nm era, highlighting where value in the AI industrial chain is concentrating. The result underscores how fabrication leaders are capturing disproportionate gains as hyperscalers and governments scramble for advanced silicon. (markets.financialcontent.com)
Taiwan Semiconductor Manufacturing did not just have a good month. On April 10, 2026, it said first-quarter revenue reached 1.134 trillion New Taiwan dollars, up 35.1% from a year earlier, and March alone jumped 45.2% to 415.19 billion New Taiwan dollars. (tsmc.com) That tells you where the money in artificial intelligence is landing right now. The biggest checks are going to the company that actually manufactures the most advanced chips designed by firms like Nvidia and Apple. (tsmc.com) (msn.com) Taiwan Semiconductor Manufacturing is a foundry, which means it runs the giant factories while other companies bring the chip blueprints. Building one of those factories costs tens of billions of dollars, so only a handful of companies on Earth can make the leading-edge chips at scale. (tsmc.com 1) (tsmc.com 2) The reason artificial intelligence pushes foundries so hard is simple: training and running large models takes huge numbers of advanced processors, and each processor needs the smallest, most power-efficient transistors available. Taiwan Semiconductor Manufacturing’s latest process family is called 2 nanometer, or N2, and the company says volume production started in the fourth quarter of 2025. (tsmc.com 1) (tsmc.com 2) This 2 nanometer shift is not just a smaller ruler mark. Taiwan Semiconductor Manufacturing says N2 uses its first generation of nanosheet transistors, a new design meant to improve performance and reduce power use compared with older 3 nanometer chips. (tsmc.com 1) (tsmc.com 2) The company is already extending that family. Its N2P version is scheduled for volume production in the second half of 2026, which means customers are not buying a one-off breakthrough but reserving spots on a moving production roadmap. (tsmc.com) That roadmap matters because artificial intelligence chips are now sold on watts as much as on raw speed. A data center with 100,000 accelerators can save enormous electricity and cooling costs if each chip does more work per unit of power. (tsmc.com) (tsmc.com) Taiwan Semiconductor Manufacturing has also been making more money from high-performance computing, the category that includes server processors and artificial intelligence accelerators. In its 2024 annual report, high-performance computing was 59% of wafer revenue, ahead of smartphones at 28%. (tsmc.com) So the 35% jump is not just a sales spike. It is a sign that the choke point in artificial intelligence is no longer only who can design the best chip, but who can secure scarce slots on the most advanced production lines. (tsmc.com) (tsmc.com) The next hard number comes on April 16, 2026, when Taiwan Semiconductor Manufacturing holds its first-quarter earnings call. That is when investors will look for two things behind the March surge: how fast 2 nanometer is ramping, and whether demand is still outrunning supply in advanced manufacturing and packaging. (tsmc.com) (tsmc.com)
