How much can Intel increase chip production for AI, according to CTO Pushkar Ranade?

Pushkar Ranade, Intel's CTO, told Dwarkesh Patel that there are not enough tools in the world for Intel to make 10x its current leading-edge wafers, but probably enough for about 2x. He says output from existing fabs grows only incrementally because yields are already high, and that unless equipment makers and their own suppliers all expand, total industry output cannot rise much.

Dwarkesh Patel noted that ASML makes fewer than 100 EUV machines a year, shared among Intel, TSMC and memory makers such as SK Hynix.
Ranade says once a process like 18A is fixed, Intel is locked into one specific tool from one supplier for each step.
Ranade estimates about 80% of the 18A tool set can be reused for 14A.

▸In their words

“Not 10x, but probably like 2x, yes.” Pushkar Ranade · at 48:42 — open in the transcript →

Source: our own transcription of the episode. The full conversation is on Dwarkesh Podcast (2026-10-22). Read the whole episode summary →

▸More from this episode

Q2Why does Intel use High-NA EUV for 14A when TSMC does not, according to Chris Auth?Q3Is Intel's chip technology behind TSMC, according to Intel CTO Pushkar Ranade?Q4What did Intel learn from its 10 nanometer delays, according to Chris Auth?Q5Will ruthenium replace copper in chip wiring, according to Intel's Jack Kavalieros?Q6How small can hybrid bonding connections get, according to Intel's Tyler Osborne?Q7What does Intel CTO Pushkar Ranade say about Elon Musk's TerraFab?Q8Does Intel regret selling its flash memory business, according to CTO Pushkar Ranade?

Ask your own recordings. Upload a podcast, a lecture or a meeting and ask it questions like this one. Transcribe with Whipscribe →