TSMC plans to raise prices across both its advanced and mature-node foundry services by up to 10 percent starting in early 2027, according to a Nikkei report relayed by chip analyst Jukan. Base price increases will range from 5 to 10 percent depending on the customer and product, covering the full range of TSMC's manufacturing lineup rather than just its most cutting-edge chips. Customers who place additional high-performance computing orders beyond their original forecasts, the kind of last-minute capacity requests common among AI chip designers racing to secure supply, will face an extra 10 to 15 percent premium stacked on top of the base increase. Price negotiations with customers reportedly began in June and wrapped up in July, meaning the new pricing is already locked in rather than still under discussion. TSMC cited rising costs for materials, semiconductor manufacturing equipment, and the construction of new overseas fabs, including its expanding plants in the United States and Japan, as reasons for the increase. As the dominant foundry for Apple, Nvidia, and AMD's most advanced chips, TSMC has consistently been able to pass higher costs on to customers who have few alternatives, and this latest round shows that pricing power still has room to grow.
Why it matters · Higher foundry prices flow straight to TSMC's margins since customers have nowhere else to go for leading-edge chips, while the steeper premium on rush AI orders shows just how much pricing power TSMC now has over Nvidia, AMD, and other chip designers scrambling for capacity.
Worth asking · TSMC is charging up to 25% more for rush AI chip orders on top of a broad 5 to 10% base hike, and customers are apparently accepting it. Does this show AI chip demand is still strong enough to absorb higher costs, or is it a warning sign for the margins of TSMC's own customers?