Intel CTO: Frontier Wafer Output Can at Most Double, EUV Tool Supply Caps AI Compute Expansion

In a recent interview on the Dwarkesh podcast, Intel Chief Technology Officer Pushkar Ranade said that even with ample funding, global output of frontier wafers is unlikely to expand tenfold, with a more realistic ceiling of around twofold. He attributed the constraint to the semiconductor supply chain's ability to scale up collectively, rather than to any single bottleneck.

Ranade noted that AI compute has grown roughly threefold per year over the past decade, but the supply of key equipment, cleanrooms, and fab capacity needed to produce advanced chips is becoming a new ceiling on capacity. ASML currently produces fewer than 100 EUV lithography machines per year, with limited room to significantly increase output by the end of this decade, while Intel, TSMC, and memory makers such as SK Hynix compete for the same tools.

He also said that equipment ordered today typically takes about 18 months to deliver, meaning new capacity is unlikely to come online until around 2028. Building new fabs does not automatically lead to more wafers, largely because equipment supply cannot keep up. To raise overall output, equipment makers, their upstream suppliers, and lower-tier sub-suppliers must all expand capacity in tandem; if any link in the chain—lithography tools, key components, materials, inspection equipment, or fab construction—falls short, frontier wafer output cannot rise sharply.