Burn Multiple
Burn multiple = net burn / net new ARR · sub-1× is great · AI startups regularly run 3-5× due to compute cost.
Text reviewed October 5, 2026
Burn multiple = net burn / net new ARR · sub-1× is great · AI startups regularly run 3-5× due to compute cost.
Basic
David Sacks popularized the burn multiple: money burned per dollar of new ARR. A 1× burn multiple means $1 spent per $1 of net new ARR. Best-in-class SaaS: 0.5-1×. Average: 1.5-2×. AI startups often run 3-5× because training runs and inference compute burn cash before revenue scales.
Deep
AI burn multiples decompose into: R&D (model training, salaries), inference COGS, sales, and GTM spend. Inference COGS grows with usage but has less up-front cost. Current figures, with dates and sources, are on the BenchGecko economy pages.
Expert
OpenAI burn multiple is positive but opaque · ChatGPT consumer losses offset by API gains.
Depending on why you're here
- ·How much money a company loses to grow $1 of new sales
- ·AI companies spend a lot early to build models
- ·Gets better as they get big
- ·Track your startup's burn multiple quarterly
- ·Watch training + inference cost as % of total burn
- ·Healthy: trending down QoQ
- ·Primary capital efficiency metric for Series B+ AI
- ·AI makes burn multiples worse early, better late
- ·Burn multiple = net burn / net new ARR
- ·AI startups: 3-5× pre-scale
AI burn multiples are 2-3× higher than SaaS at same revenue · investors accept it because the trajectory flips fast at scale.