Anthropic, OpenAI face enterprise cost-cutting as customers switch to cheaper models

Enterprises are pulling back from OpenAI and Anthropic as token spending spirals and ROI remains unclear. Lindy CEO Flo Crivello switched 100% of his 25-person AI startup's traffic from Anthropic Claude to Chinese competitor DeepSeek in mid-June, cutting costs dramatically — he described it as 'a matter of survival.' The broader pattern: founders and executives across the U.S. are reining in AI spending after the 'tokenmaxxing' era where costs per company have ballooned into the billions since ChatGPT's 2022 launch. Uber burned through its entire 2026 AI budget in four months and has since imposed $1,500/month baseline spending tiers.

Anthropic reported $47 billion in May 2026 ARR (up from ~$10 billion for all of 2025), while OpenAI's run rate sits near $25 billion (up from $13.1 billion in 2025). Both companies filed confidentially for IPOs in early June. Yet the mood has shifted: enterprise customers are now demanding ROI proof before committing further. OpenAI is weighing drastic token price cuts in anticipation of Anthropic doing the same. Microsoft, Amazon, and Google are all launching efficiency-focused model tiers as alternatives. D.A. Davidson analyst Gil Luria told CNBC the current growth rates are 'the fastest they will ever be' — a strategic window closing before customer spending rationalization.

For architects choosing between frontier and efficient models, this consolidation marks a turning point: the era of unlimited token spend is over. Expect price wars to intensify before both companies go public. Caching, multi-model routing, and model selection by task (not capability) are now central to cost management. Teams that standardized on a single frontier provider now have leverage to negotiate or migrate.