The Pulse: token spend breaks budgets – what next?

This article explores the growing concern over skyrocketing token spend in AI-powered coding tools across tech companies of various sizes. Based on interviews with engineers at 15 businesses, it reveals a 10x increase in token costs over six months, driven by heavy use of frontier models like Claude and GPT. Large companies are responding by defaulting to cheaper models, monitoring usage, and raising API budgets, while some explore open-source alternatives. The piece highlights the tension between productivity gains and unsustainable costs, and hints at possible future restrictions or model prioritization.