Podcast host Dwarkesh Patel relays that OpenAI's top 10% of researchers by usage each spent about US$1,000 a day on internal coding agents in early summer and more than US$7,000 a day by late August.

If accurate, research agents inside labs are themselves a large source of compute demand. The figures are Patel's relay, and no third party has checked them. Patel is known for long interviews with AI researchers and lab leaders, and his November 13 essay argues that AI research can be automated. Patel also writes that compute at frontier labs is growing more than fourfold a year (Dwarkesh Patel, 2026-11-13). How we checked →
Source and checks
This item ran in the brief of Nov 14, 2026, alongside 14 other items. How we checked it, the quote checks and what was still missing that day are on that issue's How we checked page.
Cite this item
SecondSource (2026). "Podcast host Dwarkesh Patel relays that OpenAI's top 10% of researchers by usage each spent about US$1,000 a day on internal coding agents in early summer and more than US$7,000 a day by late August.." https://secondsource.io/en/expert/2026-11-14-1/. Accessed access date.