Enterprise AI Shifts to Consumption Pricing Sparking Budget Overruns

Enterprise AI pricing is transitioning from fixed-seat models to consumption-based structures, creating significant financial volatility for corporate adopters. Anthropic recently shifted to a model charging $20 per month per seat with additional usage billed at API rates, while GitHub Copilot implemented similar consumption-based changes on June 1.

This shift has resulted in severe budget overruns for major firms. An Amazon project utilizing Claude Sonnet reportedly exceeded its budget by 860%, resulting in a $1.8 million bill. Similarly, Uber exhausted its entire annual AI budget within four months following a surge in the adoption of coding tools.

In response to these costs, Uber introduced a $1,500 monthly spending cap per employee for agentic coding tools. Industry experts recommend that companies implement AI gateways to act as a control plane, allowing firms to attribute requests, enforce spending limits, and route tasks to the most cost-effective models based on the complexity of the work.