AI Burn ClockIndex of the cost of retrieval by reading in AI systemsAI Burn ClockIndex of the cost of retrieval by reading in AI systemsRelease for South Carolina
Federal prime contracts whose descriptions name artificial intelligence, machine learning or language models, performed in South Carolina, total $2.0M for fiscal year 2026 to date, or … per resident. Applied to that figure, the reference factors of this index put the avoidable reading at … a year, … a day. This is a scenario computed from a published floor, not an account of the state's spending.
| Series | Value | Source and note |
|---|---|---|
| Federal contracts naming AI, FY2026 to date, place of performance | $2.0M | USAspending.gov, keyword filter in the Technical note; a floor. |
| Resident population, 2024 | … | Census Bureau, Vintage 2024 estimates. |
| State debt at end of fiscal year 2023 | … | Census Bureau, Annual Survey of State Government Finances. Shown for scale; not attributed to any cost in this index. |
| Avoidable reading, scenario, per year | … | Reference factors B1–B3 of the national release. |
Agents that read whole files to find one passage are billed for every file. A local index answers "where is it" first and the agent reads only that passage. Measured on a production codebase: 16 to 47 times less read per answer; on published coding tasks: 2.7 times fewer tokens end to end. The software is open source and runs inside your own environment; nothing leaves the building.
A one-page memo for South Carolina with these figures, their sources, and a pilot that runs on one laptop with no procurement is available on request: [email protected].
Send this release to your state CIO Opens a prefilled email with the figures and the sources; add the address.
<script src="https://aiburnclock.org/embed.js" data-state="SC"></script>National release