Dwarkesh Patel

Host — Dwarkesh Podcast Co-author — The Scaling Era

Listen — profile

0:00 / 2:48

Profile

If you want to understand how the people actually building AGI think — not their press-release talking points, but their real mental models — Dwarkesh Patel has quietly become the most important place to look. Born in Vadodara, India in 2000 and raised across North Dakota, West Virginia, Maryland, and Texas, Patel started interviewing writers and technologists in 2020 while studying computer science at the University of Texas at Austin, under the podcast’s original name, The Lunar Society. Within a few years he went, in The Economist’s words, “from nowhere to become Silicon Valley’s favourite podcaster.” Tyler Cowen calls him “the No. 1 chronicler of the AI era,” and Time named him to its 100 most influential people in AI in 2024.

What separates Patel from the crowded field of AI podcasters is preparation. Before sitting down with a researcher he reads their papers, implements exercises from their technical work, and consults other experts so he can push back in real time. That rigor is why the field’s principals keep coming back: he has hosted Ilya Sutskever, Dario Amodei, Demis Hassabis, Mark Zuckerberg, Andrej Karpathy, Satya Nadella, and Jensen Huang, often for three or four hours at a stretch. When he told Jensen Huang that China has roughly 10% of the US’s compute on a FLOPs basis, or pressed guests on scaling limits and misalignment, he demonstrated the thing most access-journalism interviews lack: enough technical knowledge to know when a talking point is misleading.

For developers, Patel matters in two ways. First, his catalog is arguably the best primary-source record of the scaling era — a searchable archive of how Sam Altman, Hassabis, Sutskever, and dozens of others reasoned about capabilities, timelines, and risk while it was happening. Second, he has become a genuine analyst in his own right. His 2025 essay arguing that AGI is not right around the corner — with continual learning as the core bottleneck — landed hard enough that both Altman and Hassabis publicly agreed continual learning is a major obstacle. He is a podcaster who moved the discourse, not just recorded it.

His book The Scaling Era, co-written with Gavin Leech and published by Stripe Press, distills years of these interviews into a single annotated volume — heavily footnoted, with definitions and diagrams — making it one of the most accessible on-ramps to how frontier researchers actually think. If you’re learning AI in 2026, Patel’s feed is closer to a graduate seminar than to entertainment.