AI Coding Tool Costs to Exceed Developer Salaries by 2028: Gartner Report

Expenditure on artificial intelligence coding tools will exceed the salary of a software developer by 2028, according to a report by Gartner. Its forecast is based on an increase in token consumption by language models and a transition to consumption-based licensing models by technology vendors.

Under consumption-based pricing, token consumption directly determines the price of coding tools. The report notes that technology vendors provide limited transparency regarding how token consumption is calculated and billed, which prevents enterprises from forecasting and controlling budgets.

Nitish Tyagi, Senior Principal Analyst at Gartner, said in a statement, “Token discipline will not emerge through developer choice alone, as developers tend to optimise for speed and convenience over cost efficiency. Without a governed engineering operating model, costs can escalate faster than the productivity gains these tools are designed to deliver.”

Tyagi further noted, “Most organisations still lack the maturity and frameworks to effectively measure cost versus business impact. Software engineering leaders are increasingly concerned as token-driven AI spend becomes harder to justify, with budgets often being depleted earlier than expected.”

The research identifies usage patterns and governance gaps as drivers of expenditure. Issues include autonomy in workflows without oversight, unedited context windows, and an absence of feedback mechanisms. Additionally, coding agents currently lack cost optimisation capabilities.

“AI coding costs will continue to rise as infrastructure investment and profitability challenges push model pricing higher,” said Tyagi. “At the same time, as more developers adopt AI tools, light users are expected to rapidly become mainstream users as familiarity and reliance increase, driving further growth in token consumption and overall spend.”

To manage budgets, Gartner recommended that engineering leaders implement operating models for tool usage. This involves establishing decision frameworks that classify tasks and aligning model selection with task complexity. The firm has advised directing routine tasks to smaller models while reserving frontier models for complex work.

It has also adivsed organisations to train developers in context engineering to lower token consumption by limiting input to relevant information. Finally, the report suggested introducing governance controls, such as token thresholds, and reviewing token usage during development cycles to locate inefficiencies.