TSMC posts 35% revenue surge
TSMC reported about a 35% revenue increase as AI demand accelerates and the foundry advances toward 2nm production. (markets.financialcontent.com) The result underscores that AI is reshaping semiconductor manufacturing — not just chip designers — and helps explain ongoing capacity tightness across the stack. (markets.financialcontent.com)
A chip designer like Nvidia draws the blueprint, but Taiwan Semiconductor Manufacturing Company is the factory that actually carves the circuits into silicon. On April 10, 2026, that factory said January-through-March revenue hit 1.134 trillion New Taiwan dollars, up 35% from a year earlier. (cnbc.com) That is an unusually clean way to see the artificial intelligence boom moving from software into physical manufacturing. March alone brought in 415.2 billion New Taiwan dollars for Taiwan Semiconductor Manufacturing Company, up 45.2% from March 2025. (reuters.com) Taiwan Semiconductor Manufacturing Company is the world’s biggest contract chipmaker, which means Apple, Nvidia, Advanced Micro Devices, and dozens of other firms can design chips without owning the multi-billion-dollar plants needed to make them. The company says it served about 465 customers and had more than 17 million 12-inch-equivalent wafers of annual capacity in 2025. (tsmc.com) Artificial intelligence chips are pushing that factory model harder than phones or laptops did because training a large model can require tens of thousands of accelerators running in parallel inside data centers. That demand has kept orders strong at the most advanced manufacturing nodes, where Taiwan Semiconductor Manufacturing Company has the biggest lead. (cnbc.com) The “2 nanometer” label is basically a generation name for a newer production process, like moving from a four-lane road to a denser highway that fits more traffic into the same space. Taiwan Semiconductor Manufacturing Company says its N2 generation is its first to use nanosheet transistors and is scheduled for volume production in the second half of 2025. (tsmc.com) A nanosheet transistor is a new switch shape inside the chip that gives engineers tighter control over power and speed. Taiwan Semiconductor Manufacturing Company says the follow-on N2P version is scheduled for volume production in the second half of 2026. (tsmc.com) That timing matters because customers do not buy “2 nanometer” as a science project; they buy it to pack more computing into a fixed power budget. In artificial intelligence servers, lower power per calculation can mean more chips per rack before heat and electricity become the bottleneck. (tsmc.com) The squeeze is not just in chip design. When one foundry is making the leading processors for smartphones, cloud servers, and artificial intelligence accelerators at the same time, every jump in demand can tighten capacity across the whole supply chain, from wafers to advanced packaging. (finance.yahoo.com) Taiwan Semiconductor Manufacturing Company has already been telling investors that this buildout will be expensive. Yahoo Finance reported that the company previously said it planned to spend up to $56 billion in 2025, which shows how much capital is required before a single extra artificial intelligence chip ships. (finance.yahoo.com) So the 35% jump is not just a good quarter for one Taiwanese manufacturer. It is a snapshot of where the artificial intelligence boom is bottlenecked right now: not in ideas, but in the number of advanced chips the world’s most important chip factory can physically produce. (reuters.com)
