AI agents read up to 47 times more than they use. This is the bill.

At the reference parameters, a team of 50 developers whose agents search by reading spends an estimated … a year on that reading, of which … would not have been read at all had a local index answered "where is it" first. That is … a month, or the fully loaded cost of … senior engineers a year, spent on reading nothing needed. Change any parameter below and this sentence is recomputed.

First release of this series; the change line begins with the next one.

Estimates, not accounts. The method is four multiplications with every factor on this page; the measured factor comes from a production codebase and from a published case study. Federal and Treasury figures further down are reference values shown as published, for scale, and are not attributed to this cost.

Table 1Estimated annual cost of retrieval by reading, by team size

Select a row to set the summary. Reading bill = developers × requests per day × working days × tokens read per request × price per million tokens. Avoidable = reading bill × avoidable share.

Reference parameters: . Adjust under Technical note, item A. Last column: avoidable cost divided by $185,000, a reference fully loaded cost of one senior engineer in the United States.

Table 2Cost of one answer, by retrieval method, measured

Three ordinary questions put to an agent on one production Next.js codebase, September 2026. "Read" is the size of the whole files behind the answer, which a grep-and-read agent loads into its context; "Index" is what the same agent read when a local index answered first.

Tokens ≈ bytes / 4. A published case study on real coding tasks reports 2.7× fewer output tokens end to end; this release uses the conservative 1 − 1/2.7 = 63 % as the avoidable share, not the 94–98 % above.

Chart 1Ten-year projection of the reading bill, two methods

Growth starts at the Gartner 2026 rate for worldwide AI spending and tapers to the long-run parameter. The shaded area is the gap: what is read and paid for without being needed.

Projection of the selected team's reading bill. Assumptions in Technical note, items A and B.

Table 3Federal contract obligations naming AI, by state of performance, FY2026 to date

Prime contracts whose descriptions name artificial intelligence, machine learning or language models. USAspending.gov, refreshed daily. A floor, not a total: AI work inside larger contracts is usually unlabelled. State debt at end of fiscal year 2023 from the Census Bureau's Annual Survey of State Government Finances, shown for scale. Each state has its own release page.

Tables 4 and 5Awarding agencies and recipients, same filter

Reference figuresShown as published, for scale

These are official series reproduced without adjustment. They size the environment in which the cost above is incurred; nothing on this page attributes them to it.

Technical noteEstimation method and parameters

Item A sets the team estimate (Table 1, Chart 1). Item B sets the world scenario (Reference figures). Every default carries its source; every control is yours. The full note, with what the index does not claim, is on the methodology page.

Table 6Organisations that published their own figure

Measured with the session score card, not estimated. An organisation appears here only with its written permission and in the form it chooses: named, or described by sector and size. To be listed, send the card and the permission line to [email protected].

Embed, data, and memosFor newsrooms, offices and teams

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

State and agency offices: a one-page memo with your figures, their sources, and a pilot that runs on one laptop with no procurement is available on request at [email protected]. Put the state or agency in the subject line.