Scenario analysis
What if high-paid engineers spent $250K a year on AI tokens?
Goldman Sachs says AI's economic impact has been "basically zero." Jensen Huang says companies aren't spending enough. We modeled three adoption scenarios to see what it would take to make the frontier AI labs profitable.
New annual token revenue, by adoption scenario
Billions of dollars. Based on U.S. software engineer population and compensation data.
Jensen's literal dream
~150K engineers at $500K+ comp
$38B
Trickle-down adoption
~1.9M senior + mid-level engineers
$100B
National-scale adoption
~4.4M all U.S. software engineers
$358B
OpenAI needs to break even
~$78B
in new token revenue, at 40% gross margin
Anthropic needs to break even
~$58B
in new token revenue, at 40% gross margin
Even Scenario 1 — only the $500K+ elite — generates enough new demand to cover roughly a third of OpenAI's losses and most of Anthropic's. Jensen doesn't need his full dream to come true. He needs about a fifth of it.