Gartner has predicted that the financial cost of AI coding tools will overtake the average software developer’s salary by 2028. This rapid escalation is driven by a massive surge in large language model token consumption, combined with an industry shift toward variable, consumption-based pricing models. Nitish Tyagi, Sr. Principal Analyst at Gartner, notes that while organizations are rapidly moving to scaled deployment, many are underestimating the fiscal reality, stating, “Organizations are rapidly moving from experimentation to scaled deployment of AI coding agents, but many are underestimating the financial impact of rising token consumption”.

The core issue stems from a lack of visibility and governance within organizations, especially since software developers naturally prioritize speed and convenience over cost efficiency. Tyagi warns that “Token discipline will not emerge through developer choice alone, as developers tend to optimize 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.” Furthermore, he states that “Most organizations 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.”

Exacerbating these budget overruns, AI coding vendors have yet to deliver mature, built-in cost optimization capabilities within their agents. This leaves enterprises to navigate complex operational failure modes, such as ungoverned autonomy in agent-driven workflows and heavily bloated context windows. Managing these internal workflows is critical because macro-level infrastructure investments and vendor profitability challenges will likely push model pricing even higher. As Tyagi points out, “AI coding costs will continue to rise as infrastructure investment and profitability challenges push model pricing higher. 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.”

Ultimately, the productivity promise of generative AI hinges entirely on operational discipline rather than raw technological capability. To protect their bottom line, Gartner recommends that organizations transition from complete developer autonomy to a structured execution framework that explicitly classifies tasks into three execution models: developer-led, developer-with-agent, and fully agent-led. By embedding strict controls—such as intelligent model routing for simpler tasks, mandatory context engineering practices, and token usage reviews into sprint retrospectives—enterprises can curb uncontrolled cost growth and align their AI investments with genuine business value.

Published on June 24, 2026