TSMC’s AI‑Fueled Quarter

Taiwan Semiconductor Manufacturing Co. opened 2026 with a record quarter as demand for the advanced chips used in artificial intelligence systems kept climbing. (tsmc.com) On April 16, TSMC said first-quarter revenue reached NT$1.134 trillion and net income hit NT$572.48 billion, up 35.1% and 58.3% from a year earlier. Diluted earnings per share rose to NT$22.08, and revenue came in at $35.9 billion in U.S. dollar terms. (tsmc.com) The company guided second-quarter revenue to $39 billion to $40.2 billion, above its first-quarter result, and kept full-year 2026 revenue growth at more than 30% in U.S. dollars. Chief Executive C.C. Wei said on the earnings call that “AI-related demand continues to be extremely robust.” (tsmc.com) (cnbc.com) TSMC is the world’s biggest contract chipmaker, which means it manufactures processors designed by customers such as Nvidia and Apple rather than selling chips under its own brand. When orders rise at TSMC, they are often a readout on how much of the global electronics industry is still spending. (cnbc.com) (tsmc.com) This quarter stood out because the usual seasonal slowdown did not bite as hard. TSMC’s first-quarter presentation showed 7-nanometer and smaller processes made up 73% of wafer revenue, and CNBC reported advanced chips accounted for about 75% of total wafer revenue in the quarter. (tsmc.com) (cnbc.com) Industry analysts now describe artificial intelligence as a business line large enough to reshape TSMC’s mix. The Next Platform reported on April 20 that AI could soon account for roughly one-third of TSMC’s revenue as hyperscalers and model builders keep buying graphics processors and custom accelerators. (nextplatform.com) That spending is also tightening other parts of the supply chain, especially memory. A Global Electronics Association report published April 13 said rising demand for high-bandwidth memory is pulling capacity away from conventional DRAM and NAND, with 82% of manufacturers reporting higher prices and 62% reporting constrained availability or longer lead times. (finance.yahoo.com) High-bandwidth memory is the fast memory stacked next to AI processors to keep them fed with data, and it uses manufacturing capacity that might otherwise go to mainstream chips. IEEE Spectrum reported in February that the collision between the usual memory cycle and the AI build-out had already pushed prices higher and could keep them elevated for years. (spectrum.ieee.org) TSMC said little in its quarterly materials about direct pricing pressure on silicon wafers, but it did point investors to broader risks around materials and supply disruptions. The company’s April 16 outlook also cited concerns tied to the Middle East conflict, including possible disruptions to energy supplies and chipmaking inputs such as helium and hydrogen. (cnbc.com) (tsmc.com) For now, the cleanest signal is still the quarter itself: TSMC beat expectations, raised no alarm about AI demand, and told investors to expect another bigger quarter right after this one. (tsmc.com) (cnbc.com)