By 2028, expenses associated with AI coding might exceed the salaries of developers.

      By 2028, the cost of AI tools that developers utilize may exceed their salaries, according to Gartner's warning. AI coding expenses are rapidly increasing, and most companies remain unaware of their spending patterns. The surge in AI coding has a price tag that is escalating quickly. Gartner projected on June 24 that by 2028, these coding costs will surpass the average developer’s salary. The reason is straightforward: every action an AI agent performs consumes tokens, with the costs continually accruing.

      Tokens represent the data processed by an AI model. With new pricing structures, a higher token usage results in a larger invoice. “Organizations are swiftly transitioning from trial phases to full-scale deployment of AI coding agents, but many are misjudging the financial consequences,” stated Nitish Tyagi, a senior principal analyst at Gartner.

      This warning comes at a peculiar time when AI coding tools are immensely popular. Engineers praise them, and managers acknowledge the genuine speed improvements they bring. However, these same tools now pose the risk of costing more than the personnel utilizing them. Gartner's message is clear: both popularity and costs are increasing simultaneously.

      A significant increase in costs is already apparent. AI coding expenses are rising from $20 or $100 a month per developer to $2,000 or even $5,000, as reported by Tyagi to The Register. This escalation is driven by a subtle shift in how these tools are sold. Previously, vendors charged a flat seat fee; now, most operate on a consumption basis. Developers pay for what an agent uses, and agents can consume a substantial amount. This transition has turned a predictable expense into a volatile one. Similar market dynamics have already seen enterprise AI costs triple despite token prices dropping.

      Consumption pricing benefits vendors as usage increases. The greater the quantity an agent writes, tests, and retries, the higher the bill becomes. An individual autonomous task can deplete tokens without a developer's awareness. When this is multiplied across an entire team, monthly invoices can inflate significantly.

      This doesn't imply that the tools are ineffective. When utilized properly, they accelerate feature delivery and relieve engineers of repetitive tasks. The concern lies in the disconnect between the potential benefits and actual costs. Currently, far too few teams are measuring this gap, and even fewer are taking action based on their findings.

      The underlying issue is a lack of visibility. Many vendors do not disclose how they measure or charge for token usage, as noted by Gartner. Consequently, companies struggle to predict expenses, leading to depleted budgets earlier than anticipated. “Most organizations still lack the maturity and frameworks needed to effectively correlate costs with business outcomes,” said Tyagi.

      Gartner warns that engineering leaders are finding it increasingly challenging to justify expenses driven by token usage. Budgets can vanish quicker than expected, and the absence of a method to connect spending to business value can lead to awkwardness in budget discussions.

      Developers typically do not oversee costs. Their focus is on efficiency, not budgeting. “Token management will not be achieved solely through developer initiative,” noted Tyagi. He cautioned that without regulations, expenses could escalate quicker than the productivity promised by the tools.

      Several factors contribute to rising costs. Autonomous agents that operate independently consume tokens freely. Excessively large context windows, where the tool processes more text than required, further inflate expenses. Additionally, teams often neglect to establish feedback loops to mitigate waste.

      The tools themselves offer little assistance in managing these costs. According to Gartner, AI coding vendors have yet to implement robust cost controls. Therefore, it falls to the buyer to exercise restraint, but many are unprepared for this responsibility.

      The user base is evolving too; as users become more accustomed to the tools, light users are transitioning to heavy usage. Gartner anticipates that model prices will increase as AI companies seek profitability, leading to higher costs amid increased usage.

      This situation has already prompted shifts in behavior. Some organizations have begun imposing limits on how much AI their employees can utilize, with the most AI-dependent firms reportedly spending around $7,500 per employee each month.

      The growing challenges have opened up a market opportunity. Database vendors are now marketing themselves as solutions to escalating AI costs, claiming they can reduce the number of requests made by coding agents. Others are advocating for an industry standards body to clarify billing processes.

      Even major players are reevaluating their positions. Microsoft has quietly scaled back on extensive Claude Code usage due to costs, and GitHub has suspended certain Copilot registrations as demand pressures the economic model. While these tools are effective, the challenge lies in the affordability of large-scale usage.

      Gartner envisions a broader market entering a new phase of growth and competition, which should eventually lead to improved cost management tools and clearer pricing structures. Currently, however, buyers are ahead of the products, rapidly scaling up with tools that were not designed for cost-effectiveness.

      Gartner's advice emphasizes the need for discipline, not withdrawal. It suggests that engineering leaders categorize tasks into three groups: developer-led, developer-with