Gartner Says AI Coding Token Costs Could Exceed Developer Pay by 2028
The analyst firm warns that rising token use, a shift to consumption pricing, and poor vendor metering could push AI coding bills past salaries if firms do not adopt governance and engineering controls.
Overview
- Gartner published its forecast on June 24 and said token-driven spend for AI coding agents may overtake an average developer’s salary by 2028 unless pricing or governance change.
- The firm found 23% of organisations already pay $200–$500 per developer per month for tokens and 6% pay more than $2,000, showing wide current variation in bills.
- Tokens are the billing units used by large language models and long context windows, image or code generation, and autonomous agent calls multiply token use and costs.
- Gartner argues the main drivers of runaway spend are the industry move from seat licenses to consumption pricing, limited vendor transparency on token metering, and weak internal governance.
- Companies are responding with quick caps, API proxies, caching, model routing and new FinOps models and Gartner recommends classifying use cases, embedding token limits, and measuring cost versus value.