Press

What the index measures. The estimated cost of AI agents finding information by reading whole files rather than retrieving the passage they need, priced from public model prices and measured ratios, with every parameter on the page. Official series from the Treasury, USAspending and the Census Bureau are reproduced for scale and are not attributed to that cost.

Why it matters. On one production codebase, an agent read 16 to 47 times less per answer when a local index answered first; on published coding tasks, 2.7 times fewer tokens end to end. Worldwide AI spending is forecast at $2.59 trillion in 2026, up 47 %. The habit scales with the spend.

Figures that may be quoted, with attribution to the AI Burn Clock and the underlying source

• US total public debt: $40.095T (Treasury, Debt to the Penny).
• Federal contracts naming AI, FY2026 to date: $427.8M; FY2025: $615.5M (USAspending.gov; filter on the technical note). Every state has a release page, for example aiburnclock.org/state/tx/.
• Reference estimate, team of 50 developers: shown live in the Summary of the national release, with the parameters beside it.

Embed and data

<script src="https://aiburnclock.org/embed.js" data-scope="world"></script>
<script src="https://aiburnclock.org/embed.js" data-state="TX"></script>

Machine-readable release: /data.json, refreshed daily at 00:00 UTC. Pipeline and templates: GitHub.

Statement

"An agent that reads a whole file to find one function is not thinking. It is paying. The index shows the bill, the sources, and the controls. Move a control and argue with a factor, not with us."

Attribution: Serhii Nikolaichuk, Austin, Texas; co-author of an IETF draft on attestation results; maintainer of the AI Burn Clock.

Contact

hello@aiburnclock.org. Interviews and state or agency memos on request.