AI Summary
5 min readIn a recent solo episode of the Dwarkesh Podcast, host Dwarkesh Patel lays out a stark economic tension at the heart of the AI industry: the leading labs are growing revenue far faster than they can grow their compute capacity. If current trends hold, the price of compute—the raw material of AI—could increase by an order of magnitude or more, reshaping who can afford to build and use frontier models.
The Revenue-Compute Gap
Patel begins with a striking data point: Anthropic’s revenue has grown roughly 10x year-over-year for the last three years. If that continues, the lab would need to generate between $100 billion and $150 billion in revenue this year, and nearly $1 trillion the year after. Meanwhile, the total compute available to frontier labs grows at only about 3x per year. This creates an arithmetic problem.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Dwarkesh Podcast
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:04) **The Core Tension: Revenue vs. Compute Growth** - Introduces the central puzzle: Anthropic's revenue has 10x'd year-over-year for three consecutive years, while lab compute only 3x's year-over-year.
- 2 (02:05) **The Two Remaining Options: Margins vs. Compute Price** - Argues that labs need most compute for training, leaving only two ways to bridge the revenue-compute gap.
- 3 (03:41) **Case Study: Google's Premium Compute Deal with SpaceX** - Uses a real-world example to illustrate the premium labs pay for compute.
- 4 (04:06) **The Core Mechanism: Smarter Models Monetize Compute Better** - Lays out the key argument: as AI models get smarter, they generate more value per unit of compute.
- 5 (05:20) **Implication 1: Harder for Competitors to Compete** - Explains the first major consequence of rising compute prices.
- 6 (05:38) **Implication 2: The Alchian-Allen Effect and Model Premiums** - Explains the second, most interesting implication: the best model will command a huge price premium.
- 7 (06:21) **Implication 3: Current AI Applications Get Priced Out** - Explains the third consequence: cheap AI applications will disappear.
+ Full timestamped outline available in the app
Guests on this episode
Show Notes
This is a video recording of a post I wrote last week. If you want to read the original you can check it out here.
Thanks to Mercury for sponsoring this video. Mercury’s built-in AI, Command, helps me close my books and saves me a bunch of time. At the end of each month, Command categorizes my transactions and provides its rationale for every choice: I just review, fix anything that’s off, and approve... and then Mercury syncs everything to QuickBooks. Get started at mercury.com/command
Get full access to Dwarkesh Podcast at www.dwarkesh.com/subscribe
More from this podcast
Dwarkesh Podcast →