Circle publishes a monthly examination report for USDC; this teardown walks the May 2026 document — issued 2026-06-29, reserve figures as of 2026-05-29 — as parsed and hash-anchored on the USDC 2026-05 attestation page. Figures are the document's own, at its stated scope. Nothing here is a judgment of any issuer's solvency.
What stands out in May is that the composition moved more than the headlines did. Reserves and circulation contracted together by about 1.16 billion dollars — near-lockstep, as the steady ratio implies — but inside the reserve, roughly 1.8 billion left overnight repo while direct Treasury holdings grew by about 3.0 billion. A reader who only checks the coverage ratio would call this an uneventful month; the interesting part is a size-adjusted reshuffling of what the reserve actually holds.
The headline figures
| Figure | Amount (USD) | As of |
|---|---|---|
| Total USDC reserve assets | 75,961,621,872 | 2026-05-29 |
| USDC in circulation | 75,885,403,148 (tokens) | 2026-05-29 |
| Coverage ratio (stated total over stated circulation) | 1.001004 | 2026-05-29 |
The report states two as-of dates (2026-05-05 and 2026-05-29); the composition above is at the later one. Publication lag — issue date minus latest as-of date — is 31 days, measured from the document's own dates.
Composition: where the reserve sits
| Category (document's own labels, abridged) | Amount (USD) |
|---|---|
| U.S. Treasury repurchase agreements | 45,171,000,000 |
| Total U.S. Treasury securities | 22,061,994,431 |
| Cash held at regulated financial institutions | 10,794,851,072 |
| Cash held in Circle Reserve Fund | 1,003,103,452 |
| Timing and settlement differences, net (two lines) | −446,782 and −3,068,880,301 |
Beyond the category rows, the document itemizes 30 individual positions with instrument identifiers — 46 parsed rows in total, cross-footed before anything landed.
Month over month: out of repo, into Treasuries
Against the April 2026 report (as of 2026-04-30):
| Line | 2026-04 | 2026-05 | Delta |
|---|---|---|---|
| Total reserve assets | 77,123,668,726 | 75,961,621,872 | −1,162,046,854 |
| USDC in circulation | 77,047,590,794 | 75,885,403,148 | −1,162,187,646 |
| Treasury repurchase agreements | 46,998,000,000 | 45,171,000,000 | −1,827,000,000 |
| U.S. Treasury securities | 19,110,672,894 | 22,061,994,431 | +2,951,321,537 |
Circulation and reserves both contracted by about 1.16 billion dollars, and within the reserve roughly 1.8 billion rotated out of overnight repo while direct Treasury holdings grew by about 3.0 billion.
The arithmetic above is the parsed document pair and nothing more: repo down about 1.8 billion, direct Treasuries up about 3.0 billion, on a smaller total. Interpretation — market color, clearly separated from that arithmetic: moving out of overnight repo into held Treasury securities trades daily-resetting exposure for locked-in short-dated yield, a modest extension of duration posture. That is the shape a manager chooses when the front end looks more attractive to hold than to roll. The document states no motive, so none is inferred here — the rotation is worth flagging mostly as a thing to watch for continuation, because one month is a data point and two is a posture.
What the document's metadata states — and does not
- Engagement type: examined, under AICPA attestation standards, per the document's own citation.
- Accountant name: no accountant name appears as a text span in the captured PDF — recorded as not found in the document, not inferred from outside it.
- Per-chain circulation: stated as one total; no per-chain breakdown appears, so no per-chain reconciliation is derived.
- Disclosure score: 9 of 11 rubric dimensions met (the two above are the unmet ones) under rubric tb21-disclosure-v1.
One housekeeping note: this month's row was re-parsed under an upgraded template, so the attestation page shows a restatement flag — the superseded parse remains on the evidence chain, independently verifiable, with the same captured source hash.
For the June 2026 report, three things are worth watching. First, whether the repo-to-Treasuries rotation continues — a second month would make it a posture rather than a data point. Second, whether circulation resumes growth after May's 1.16 billion contraction, and whether reserves track it as closely as they did here. Third, on the disclosure side, whether either unmet rubric dimension changes — an accountant name appearing as a text span in the document, or circulation itemized per chain. The score moves only when the documents do, which is exactly how it should be.