AI Token Costs Could Exceed Developer Salaries Within Two Years, Gartner Warns

Gartner has warned that enterprise AI token costs for developers could meet or exceed their monthly salaries within two years, based on a global average salary of $2,000. The prediction highlights a rapid escalation in generative AI and agentic tool adoption, coupled with consumption-based licensing models that vendors are adopting to recover infrastructure investments.

Senior principal analyst Nitish Tyagi said he has heard organizations report individual developer token consumption as high as $20,000 in a month, and business users burning $32,000. The goal of the prediction is to alarm the industry about the impact of uncontrolled token spend, Tyagi told InfoWorld.

Enterprise Governance Struggles to Keep Pace

AI coding vendors have yet to deliver mature, built-in cost optimization capabilities. Agent-driven workflows are difficult to govern, context windows bloat, budgets are wiped out early, and token spend becomes hard to justify. Without a governed engineering operating model, costs can escalate faster than productivity gains, Tyagi said. Many organizations lack the frameworks to determine ROI as AI moves too fast.

Princeton University’s CEO-Bench test reinforces these concerns. Researchers simulated 500 days of running a software company with AI agents. Only three models finished above starting capital. Most went broke, and a simple rule-based heuristic without AI outperformed nearly all models.

Infrastructure Costs Ripple Across Industries

Big Tech’s $8 trillion AI bet is driving a massive industrial expansion that raises costs for consoles, cars, and electricity. Data centers require dense clusters of advanced chips, cooling systems, fiber networks, and backup power. Columbia University economist Stijn Van Nieuwerburgh described the effort as straining resources beyond the tech sector.

Compute is becoming the tech industry’s scarcest commodity. In March, Google told Meta it could not supply all the Gemini capacity Meta wanted to buy, disrupting and delaying internal AI projects. Meanwhile, SoftBank’s Masayoshi Son questioned Elon Musk’s orbital data center plans, noting electricity is only 7% of costs, and the AI race will be won on Earth within a few years.

What Comes Next

Enterprises must implement cost governance now or face budget overruns that undermine AI’s value proposition. Vendors will need to offer transparent pricing and optimization tools. Simpler, non-AI heuristics may prove more effective for some automation tasks. Without restraint, token costs could become a barrier to AI adoption itself.