Infineon, TI face 30‑week lead times
- Power semiconductor lead times have stretched to about 30 weeks at integrated device makers including Infineon and Texas Instruments, as AI server demand rises. - TrendForce said power-management chip lead times for general servers are moving from 21–26 weeks toward 35–40 weeks as AI orders get priority. - Infineon and TI are expanding AI power capacity while server suppliers warn bottlenecks could spill into 2027. (trendforce.com)
Power chips are becoming the next supply bottleneck in artificial intelligence servers, with lead times stretching to roughly 30 weeks at some major suppliers. (semimedia.cc) The squeeze is hitting integrated device manufacturers including Infineon Technologies and Texas Instruments, according to industry sources cited Monday by SemiMedia. The report said shortages are emerging in power components as AI data centers pull more supply. (semimedia.cc) These are not the headline chips like graphics processors. They are the semiconductors that convert and regulate electricity inside servers, power shelves and racks, and AI systems need more of them because they draw far more power than standard servers. (semimedia.cc) (trendforce.com) That demand is also changing the way data centers move electricity. SemiMedia said suppliers are rolling out 800-volt direct-current designs to cut transmission losses, while Infineon has said older 54-volt setups are nearing their limits for AI workloads. (semimedia.cc) (infineon.com) TrendForce reported on April 15 that suppliers are prioritizing higher-value AI server parts over general server components. It said power-management integrated circuit lead times are expected to extend from 21–26 weeks to 35–40 weeks as 8-inch Bipolar-CMOS-DMOS capacity is redirected to AI products. (trendforce.com) The research firm also said AI server shipments are still projected to grow about 28% in 2026, even as component delays weigh on the broader server market. Its forecast for total server shipment growth this year was cut to 13% from a level that had been expected to approach 20%. (trendforce.com) Infineon has already moved to add capacity. On February 4, the company said it would raise planned fiscal 2026 investments to about €2.7 billion from €2.2 billion to accelerate manufacturing for AI data-center power supplies, and it said this business could reach about €2.5 billion in revenue in fiscal 2027. (infineon.com) Texas Instruments is also seeing the pull from data centers show up in results. TI reported April 22 revenue of $4.83 billion, up 19% from a year earlier, and Chief Executive Haviv Ilan said growth was led by industrial and data center demand. (ti.com) Onsemi is part of the same power-chip buildout, though its next earnings report is scheduled for May 4. The company has separately said it is building out technology for the full AI data-center power chain, from the grid down to processor core voltage. (investor.onsemi.com) (marketscreener.com) The result is a familiar semiconductor pattern in a new place: AI demand is no longer straining only graphics processors and advanced packaging. It is now tightening the market for the chips that keep those systems powered, cooled and stable. (semimedia.cc) (trendforce.com)
