Capex payback gap
Whether AI revenue grows fast enough to pay for the buildout. An open test, drawn as a brake because failing it cuts capex.
brakeEmerging Open on the map →
Facts
- Lane
- Capital and talent
- Wave
- 6 of 12 (build order, not a date)
- Needs
- 11 boxes all the way back through hard links
- Unlocks
- 0 boxes downstream through hard links
- Cluster
- Around datacentre capex
Evidence as of October 2026
Sequoia’s David Cahn puts the revenue needed at about $1.5tn a year, and Bain at around $6tn. OpenAI and Anthropic’s combined run rate is about $105bn, and falling prices per unit of capability make the gap harder to close.
- 01David Cahn: AI’s $1.5T question
- 02CIO: $6T in annual AI revenue needed to pay off data centres (Bain)
- 03Epoch AI: An update on AI’s most important number
Watch
Revenue against capex: Cahn’s $1.5tn a year needed, against about $105bn of frontier-lab run rate.
Detail boxes 1
How many years a GPU is depreciated over and what it resells for, which sets reported hyperscaler profits and the value of GPU collateral.
Alphabet and Microsoft depreciate servers over 6 years, up from 3 in 2020, while a three-year-old H100 system resells for about 45% of the price of a new one (Silicon Data via Dow Jones, 8 Dec 2025).
Around it
Nothing further on the map.