Gartner warns AI coding costs will outpace developer salaries by 2028

Gartner analysts predict that AI coding expenses could exceed the average software developer’s salary by 2028. The surge is driven by rising token consumption in large language models and a shift from seat‑based licences to consumption‑based pricing, which makes costs highly variable and often opaque. “Organizations are rapidly moving from experimentation to scaled deployment of AI coding agents, but many are underestimating the financial impact of rising token consumption,” said Nitish Tyagi, senior principal analyst at Gartner. He added that token discipline will not emerge from developer choice alone, as developers prioritize speed over cost efficiency.

Many vendors do not disclose how tokens are measured or billed, leaving enterprises unable to forecast spend and risking budget overruns. Gartner recommends a disciplined operating model: define use‑cases, match model size to task complexity, enforce context‑engineering practices, and implement governance to monitor token usage. Some firms are adopting “FinOps for AI” to bring financial accountability to AI spend.

A real‑world example comes from Uber, which exhausted its AI‑coding budget within months, highlighting how unstructured token use can rapidly drain resources. The company’s experience underscores the need for task‑level model routing—using smaller, cheaper models for routine tasks and reserving frontier models for complex work—to cut spend by up to 70% without sacrificing productivity.