AI’s $600B Question
Source: https://www.sequoiacap.com/article/ais-600b-question/ Author: David Cahn (Sequoia Capital) Date: 2024-07-21
Summary
Sequoia’s analysis asking whether AI revenue can justify the enormous GPU infrastructure investment — specifically, whether there’s a $600B revenue gap between current AI revenues and the investment being made in infrastructure.
Key Claims
- Infrastructure investment: NVIDIA alone is generating ~$600B in annualized GPU revenue; data center capex is even higher.
- The gap: current AI software revenue is in the tens of billions — far less than the infrastructure investment.
- The bull case: this is like the early internet — infrastructure builds ahead of the revenue, which arrives later.
- The bear case: AI may never generate enough software revenue to justify the infrastructure. A significant write-down may occur.
- Where the revenue needs to come from: enterprise software automation is the most credible large market.
Connection to Other Sources
Connects to Token Economy Bubble — both analyze whether AI revenue justifies AI investment.
Concepts
- Scaling & Compute — compute investment vs. revenue analysis