AI Subscription Pricing Bubble: Enterprise Costs Set to Skyrocket
Big AI companies like OpenAI, Anthropic, Google, and xAI are currently engaging in unusual pricing strategies, offering enterprise AI subscriptions at significantly lower prices than it costs to run them. OpenAI's VP of Product, Nick Turley, mentioned that their $20 monthly fee was somewhat accidental. For Anthropic, users are burning through over $8 in computing power for every $1 earned. Microsoft has reportedly lost more than $20 monthly per GitHub Copilot user. As AI systems become more efficient, usage rates are skyrocketing. Starting June 1, 2026, GitHub will switch to a pay-per-use model. OpenAI anticipates a staggering $115 billion cash burn by 2029. Both OpenAI and Anthropic are eyeing IPOs, which could mean price hikes ahead.
Key facts
- AI labs are selling enterprise subscriptions below cost as a loss-leader strategy.
- Anthropic users consume $8 in compute for every $1 of subscription revenue.
- Microsoft lost over $20 per user per month on GitHub Copilot.
- GitHub moves to usage-based billing on June 1, 2026.
- OpenAI projects $115 billion cumulative cash burn through 2029.
- OpenAI committed $665 billion in compute spending by 2030.
- KPMG reports U.S. organizations project average AI spending of $207 million over 12 months.
- Goldman Sachs finds many large companies overrunning AI budgets by orders of magnitude.
Entities
Institutions
- OpenAI
- Anthropic
- Meta
- xAI
- Microsoft
- GitHub
- KPMG
- Goldman Sachs
- Oracle
- Conga
- Outlever
- The Wall Street Journal
- University of Chicago
- Marketplace
Locations
- United States