Intel's CTO warns of supply chain limits as AI chip production faces capacity constraints

In a recent interview with Dwarkesh podcast, Intel's Chief Technology Officer Pushkar Ranade highlighted the challenges in scaling advanced wafer production. He stated that even with ample funding, global production of cutting-edge wafers is unlikely to increase tenfold, with a more realistic growth cap of around double. Ranade attributed this limitation to the collective capacity of the entire semiconductor supply chain, rather than any single component. Ranade noted that AI computing power has grown approximately threefold annually over the past decade, but the production of advanced chips is now constrained by key equipment, cleanrooms, and foundry capacity. ASML, for instance, produces fewer than 100 EUV photolithography machines annually, and significant expansion is unlikely by the end of this decade. These machines are in high demand by companies like Intel, TSMC, and SK Hynix. He explained that orders placed today typically take 18 months to deliver, meaning new capacity will not be available until around 2028. New foundries also cannot automatically increase output due to the slow supply of equipment. Ranade emphasized that all suppliers, from equipment makers to raw material providers, must scale simultaneously to boost wafer production significantly.