TokenBudgeting:
SemiAnalysis Twitter · SemiAnalysis (@SemiAnalysis_) · 2026-06-30
SemiAnalysis surveyed over 50 enterprise customers and found that media reports of AI token budget blowouts are overstated, with 90th-percentile customers driving most API revenue while the median Fortune 500 employee still spends under $100 per year on AI.
Appears in
Extraction
Topics: enterprise-ai-adoptiontoken-economicsai-spendingai-market-dynamics
Claims
- Widely reported tokenmaxxing budget blowouts at companies like Meta and Uber are overstated and stem from poor incentives not found at most organizations.
- 90th-percentile and above customers account for most AI API revenue and are at little risk of cutting spend through the rest of 2026.
- Enterprise AI budgets now vary widely, from $250 to tens of thousands per employee per month, with no consensus number.
- Over 70% of ARR at OpenAI and Anthropic can be attributed to coding use cases.
- The median Fortune 500 employee spends well under $100 per year on AI, indicating the enterprise S-curve has significant remaining runway.
Key quotes
Widely reported responses to Tokenmaxxing budgets from companies like Meta and Uber are overstated and stem from poor incentives and employee allocation we didn't find present at other organizations.
We believe that over 70% of ARR today across OpenAI and Anthropic can be attributed to coding use cases with Anthropic spend higher than OpenAI's given the difference in B2B vs B2C mix.
What the coding market has done to AI Lab ARR will be repeated with Cyber (Mythos re-release dependent) at an even faster pace than Claude Code and again with white-collar knowledge work.