Research by Gartner Peer Insights has found that 23% of tech leaders are spending $200 to $500 per developer per month on tokens for artificial intelligence (AI) coding agents, such as Claude Code, Cursor and OpenAI Codex. With many organisations ramping up the use of AI to augment software engineering workflows, Gartner believes they may find their software development costs skyrocket, as AI tools providers switch to token-based pricing.
Gartner warns that the shift by AI coding agent providers from seat-based licensing to consumption-based pricing is introducing highly variable cost structures for software engineering workloads. It said that many providers of AI coding agents lack transparency into how token consumption is calculated and billed, limiting companies’ ability to accurately forecast and control costs.
Without clear visibility into token usage across development tasks, organisations risk budget overruns and reduced ability to track cost-to-value outcome.
The IT analyst firm has forecast that by 2028, AI coding costs will overtake the average developer’s salary due to rising large language model (LLM) token consumption and the shift to consumption-based licensing models.
Nitish Tyagi, associate principal analyst in Gartner’s software engineering practice, said: “I’m a big believer that AI is bringing gains. You should not move away from AI because the total costs are increasing. But I believe that token costs will certainly increase.”
As AI tools providers move to a consumption-based pricing model, he said IT leaders will see the cost of software development increase tenfold, or even 100-fold. “Cost will increase from $20 to $200, or even, in some cases, to $2,000 per month,” he warned, which is more than the average salary of a software developer based in India.
In fact, Gartner’s research shows that 6% of organisations are paying more than $2,000 per developer per month, based on token pricing in AI tools.
Tyagi advises IT leadership teams to justify the value of paying large monthly fees to support the use of AI for internal software development. As an example of where costs can be justified, he said at one Indian IT organisation, a developer’s AI usage was costing the business $20,000 based on the volume of tokens being consumed. But when the IT leadership team looked into the high cost, it found that the developer was actually working on a legacy modernisation project, and the cost could therefore be justified.
In its How to optimize token consumption for AI coding paper, Gartner urges software engineering leaders to work with their teams to assess the maturity of their current software engineering processes. The goal is to identify whether the organisation is ready for autonomous agentic development. After assessing existing software development processes, Gartner recommends that software engineering team leaders should then identify how and where developers are currently using AI coding agents.
The report authors said the goal is to identify the use cases where autonomous development and premium models are essential versus where the cost of using them is not justifiable.
