Stop Counting Tokens, Start Counting What Ships
Why rationing agentic coding backfiresβand what to measure instead.
By Affan Syed | November 2, 2026 | Artificial Intelligence | Sponsored by HPE
When a developer hands a coding task to an AI agent, the agent doesn't just answer once. It plans, generates code, runs tests, and iterates to correct its own errors. A single task can consume 10 to 50 times more tokens than a basic chat prompt. And the more useful the agent becomes, the more calls it makes. Under per-token pricing, the very workflows that create the most value can also become the most expensive.
The financial pressure is already showing. Uber reportedly burned through its entire 2026 AI budget by April after expanding its use of Claude Code. GitHub Copilot shifted from flat-rate pricing to metered credits. And Gartner predicts that AI coding token costs will surpass the average developer salary by 2028 as token consumption surges.
Sponsored content provided by HPE.