Tether Q2 2026: a $1.5B profit and a $4.2B loss

Tether headlined its second quarter of 2026 with $1.5 billion in net operating profit. However, the signed report published the same day places the financial result for that quarter at a loss of approximately $4.211 billion. Both figures originate from documents released simultaneously: the first excludes changes in asset value, while the second incorporates them. Between the two lies the explanation for why the reserve cushion—the excess capital above the 1:1 backing of every USD₮ in circulation—dropped from $8.232 billion to $4.110 billion in ninety days. We reconstruct the full bridge using both assured reports, debunk the widely reported drop in U.S. debt exposure from $141 billion to $115 billion currently circulating, and correct our own analysis from July 19.

Why does the most profitable quarter coincide with the year's lowest cushion?

The two series Tether chose to headline point in opposite directions. Operating profit rose from $1.04 billion in the first quarter to $1.5 billion in the second: a 44% increase, a figure several media outlets rounded up to 50%. Excess reserves took the opposite path, falling from $8.232 billion to $4.110 billion, a 50.1% decrease. Tether closed 2025 with a $6.338 billion cushion, climbed to a record high in March, and by June fell below the year's starting point.

The detail that organizes the rest of the analysis is found in the balance sheet. Liabilities barely moved: $183.536 billion on March 31, $183.642 billion on June 30—a difference of just over $100 million. Total assets fell from $191.768 billion to $187.751 billion. The cushion compressed because the reserve lost value while liabilities remained virtually frozen, without a corresponding issuance of USD₮ against the same backing.

The first two rows explain the quarter's tension and should be clearly distinguished. The "financial result" is a line from the assured report itself, within the statement of changes in net equity; "net operating profit" is the metric Tether selects for its press release. In the first quarter, both matched at $1.04 billion. In the second, they diverged by $5.711 billion.

The table also clarifies the two USD₮ figures that coexist in the same document and are often cited interchangeably. The gross redemption value of all tokens issued is $184.589 billion; the accounting liability for tokens is $183.622 billion. The difference consists of $808 million in tokens held by the company outside its treasury wallet, plus the adjustment for the 10 basis point redemption fee.

Where did the $4.122 billion lost from the cushion go?

The June report does not publish a breakdown for the standalone quarter because its statement of changes in equity covers the entire first half of the year: it starts from $6.338 billion on December 31, 2025, applies a financial result of −$3.171 billion and net capital movements of +$943 million, reaching $4.110 billion in June. The March report publishes the same table for the first quarter: $6.338 billion starting point, financial result of +$1.04 billion, capital movements of +$854 million, closing at $8.232 billion. Subtracting the second from the first leaves the isolated quarter, and there is no residual discrepancy.

The two middle rows are CleanSky arithmetic based on lines declared in the two assured reports; the opening and closing figures are published as-is. The bridge balances to the million. What the report does not break down is the internal composition of that $4.211 billion negative result, and for that, the asset detail—which both documents publish category by category—is essential.

Gold and Bitcoin explain most of the gap, and the reports provide closing prices to calculate this without estimates. Tether held 132.2 tons of gold at $4,668.06 per ounce on March 31 and 146.2 tons at $4,008.02 on June 30 (Bloomberg prices cited in each document): applying that drop to the tons already on the balance sheet yields approximately $2.805 billion in unrealized losses. The 97,137 Bitcoins from March, valued at $68,193.95, were worth $58,642.15 in June, roughly $928 million less. The 14 tons and 1,796 Bitcoins purchased during the quarter do not affect the cushion, as they represent a swap of cash for assets within the same reserve.

With these two pieces factored into the −$4.211 billion financial result and the declared +$1.5 billion operating profit, there remains approximately $1.978 billion in losses that neither document attributes to a specific asset. The candidates are in the table: publicly traded equities, which the report defines as indirect exposure to gold, Bitcoin, and other assets, and other investments, whose balance mixes purchases with price variations without separating them. The operational conclusion remains the same: the entire quarter's profit was not even enough to offset the loss in gold value.

Did Tether's U.S. debt exposure drop from $141 billion to $115 billion?

No. That comparison pits two different perimeters against each other, and both reports allow for line-by-line verification. The $141 billion Tether claimed in May was not its direct T-bill portfolio: it corresponded to the cash equivalents subtotal from the March report, which sums U.S. Treasury bills, overnight reverse repo, term repo, and bank deposits. That subtotal was $141.223 billion. The approximately $115 billion currently circulating is something else: the direct Treasury bills line from the June report, $114.961 billion.

When compared using the same criteria, the numbers look like this: Direct and indirect exposure to U.S. debt—bills plus both types of reverse repo, all collateralized with Treasuries—went from $141.116 billion to $140.580 billion: a decrease of $536 million, or 0.4%. The direct T-bill portfolio did drop by $2.075 billion (1.8%), but term repo absorbed much of that shift with an increase of $2.248 billion. This is a recomposition within the liquid segment with the overall size remaining almost intact.

There is a detail in the fine print of this recomposition. The March report described reverse repos as operations "where the ultimate issuer or guarantor has a rating of A-2"; the June report says "at least A-3," the lowest rung of investment grade on the short-term scale. The declared floor for counterparty quality dropped one notch in the same quarter that term repo grew by 47%. These operations are fully collateralized with Treasury debt, so the risk is buffered, but the change appears in the assured document and in no press release.

Regarding the ranking of foreign holders, it is worth remembering what that table measures. The U.S. Treasury publishes monthly, within the TIC system (Treasury International Capital, its record of international capital), a list of major holders of public debt by country of custody. It does not name companies: any position assigned to Tether is an inference made by crossing its declared exposure with country-level figures. The latest official data is from May, published on July 14, one month before the quarter's close; the release with June data arrives on August 17.

Using the broad perimeter, Tether remains wedged between Brazil and Saudi Arabia, the 17th position it claimed in May, although the margin over the Saudis has narrowed to less than $300 million. Using the narrow perimeter, it would fall below Israel, four spots lower among the holders in the table. The fact that the choice of criteria shifts the picture so significantly explains why the company stopped publishing it: in March, the round number was a brand argument; in July, the press release contains neither the amount nor the rank. What the documents do support is that sovereign exposure barely moved and that the "collapse" narrative stems from comparing a subtotal to one of its component lines.

The underlying movement of the month helps calibrate the scale: in May 2026, foreign residents reduced their Treasury bill holdings by $43.5 billion, and official foreign holdings fell from $457.3 billion to $396.2 billion in a single month. Against that, Tether's $536 million is noise.

What does this attestation correct from our July 19 publication?

On July 19, we published an analysis of Circle and Tether's exposure to Fed rate cuts that described the $8.232 billion cushion as an "all-time high" and concluded that even the harshest scenario—a cumulative cut of 125 basis points, with an estimated cost of about $1.590 billion per year in lower yield—would leave Tether "comfortably in the positive and with its cushion intact." That reading was correct based on the Q1 data available then, but it ceased to be so twelve days later.

The math changes as follows: With an $8.232 billion cushion, the 125 basis point scenario represented over five years of yield loss before exhaustion, ignoring any other income. With $4.110 billion, it represents about two and a half years. The fundamental correction is one of origin rather than magnitude: half the cushion vanished without the Fed moving rates, due to the price of gold and Bitcoin. The rate sensitivity we analyzed in July measures slow risk; the fast risk turned out to be in macro asset prices, acting upon a cushion that is now half as thick.

What does a KPMG audit add that a BDO attestation does not cover?

It is worth first specifying what the existing document actually is, as its value is often unfairly dismissed. The June report is a reasonable assurance engagement under the ISAE 3000R standard—the high end of the scale—signed on July 31, 2026, by BDO Advisory Services in Milan. BDO obtained confirmation letters from banks and custodians, performed inventory and quality tests on physical gold, verified reconciliations between accounting and on-chain ledgers, and reviewed a sample of secured loans and their collateral. It is not a light rubber stamp.

Even so, the distance from a full audit of financial statements is real, and the report itself outlines it:

• It is a one-day snapshot. BDO expressly limits its opinion to June 30, 2026, and warns that it did not apply procedures to any other date or time.
• Notes are outside the scope. The report states that the notes to the document are provided by the board "for information purposes only" and that BDO expresses no opinion on them.
• No assurance on going concern. The board applies the going concern criterion, which requires judgment on liquidity, market, and credit risks; BDO explicitly states it provides no guarantee on that assessment.
• The perimeter is a single entity, not the group. The report covers Tether International, S.A. de C.V. The group's proprietary investments remain outside the token reserves, and that boundary is defined by the company itself.
• Valuation assumes normal markets. BDO specifies that it does not reflect extraordinary conditions or cases of custodians or counterparties with substantial illiquidity, and that management identified no provision for expected credit losses.
• There is unprovisioned litigation. Both reports record a class-action lawsuit in New York courts regarding the drop in Bitcoin prices in 2017 and 2018, the outcome of which the company says it cannot estimate and for which it recognizes no provision.

What KPMG would add is what none of those lines cover: an opinion on internal controls under PCAOB standards, a breakdown of related-party transactions, and a time series rather than a single date. Tether hired them in March 2026, with PwC preparing internal systems, according to CoinDesk. The Q2 press release dismisses this in one sentence—"during the quarter, the audit process with the Big Four continued"—without naming the firm, providing a date, or defining the scope.

The regulatory calendar puts a clock on the engagement. The GENIUS Act, signed on July 18, 2025, requires issuers with more than $50 billion in circulation to publish annual financial statements audited under PCAOB standards, and sets July 18, 2028, as the day digital asset service providers will no longer be able to offer U.S. users stablecoins from non-authorized or non-qualified foreign issuers. With 184.589 billion USD₮ issued, Tether is nearly four times above the threshold. What we wrote in the comparison between USDC, USDT, and decentralized alternatives—that a Big Four audit would eliminate the historical transparency objection—remains true, with the caveat that the milestone only arrives when KPMG signs. For the practical difference between attestation and audit, the msUSD attestation crisis serves as a contrast.

What does the Federal Reserve say about the quality of USD₮ reserves?

The "Stablecoins in 2025" note published by the Federal Reserve on April 8, 2026, makes a distinction that the June breakdown allows us to recalculate. The Fed estimated that USD₮ maintained around $1.04 in reserves per coin in circulation, but only about $0.74 in assets it classifies as high quality: Treasuries, Treasury-backed repos, and bank deposits. Gold and Bitcoin do not fall into this category. USDC, according to the same note, maintained a full 1.0 backing with high-quality reserves.

Recalculating this with the assured reports, the high-quality segment amounted to $141.223 billion in March and $140.620 billion in June, against token liabilities of $183.438 billion and $183.622 billion. The result is $0.77 per coin in both quarters: the reserve composition did not degrade; rather, the value of the segment already outside that category changed. Overcollateralization, however, was cut in half, from 4.5% of token liabilities to 2.2%, and the 1.04 multiplier approached 1.02. Since the GENIUS Act defines admissible reserves as dollars and short-term public debt, the assets that sank the cushion are precisely those that fall outside both definitions.

What do 30 million new users mean with half the cushion?

The Q2 press release states that the global user base grew by more than 30 million during the quarter and places USD₮'s market share above 60% of the stablecoin market. Issued USD₮ grew by 446 million, a small increase relative to the base size: most of these users are entering existing stock. This is the pattern of a currency used as payment infrastructure where access to banking dollars is scarce, a phenomenon we analyzed in the stablecoin payments market radiograph.

The intersection with the cushion changes the scale of risk for the end user. In March, every $1,000 in USD₮ was backed by about $45 of excess capital; in June, by about $22. This capital is what absorbs reserve losses without affecting the peg. For those maintaining operational balances and holding their own keys, the practical conclusion involves diversification across issuers and networks, a topic we covered in the USD₮ multi-chain self-custody guide. In the European Union, the question is resolved differently: Tether does not appear in ESMA's register of electronic money tokens, as detailed in our review of the MiCA EMT register.

What to look for in the third quarter attestation?

The Q3 2026 attestation should be published in late October, and there are five lines worth checking first:

• The statement of changes in equity. It will cover the first nine months, so the standalone quarter will again be derived by subtraction. If the financial result returns to positive, the cushion rebuilds; if it repeats as negative, two consecutive quarters cease to be a mere pricing episode.
• The repo rating floor. If the wording remains "at least A-3" and term repo continues to grow, the shift from March to June is no longer a one-off.
• The weight of gold and Bitcoin. Tether added 14 tons of gold in the quarter where the metal cost it $2.8 billion in cushion, pointing to an allocation maintained by conviction.
• A date for KPMG. With the July 2028 threshold on the horizon, the room for the process to remain described in a single sentence is narrowing.
• The detail on equities and other investments. These are the two categories absorbing the $1.978 billion that currently cannot be attributed to a specific asset.

Before that, the August 17 TIC release with June data will allow for the sovereign comparison to be placed in the exact month of the quarter's close. The two figures that opened this article coexist without contradiction: Tether earned $1.5 billion operating and lost $4.211 billion in the same quarter, and both are its own. The first is what went to the headline; the second is what explains the cushion.

Sources and links: Tether/BDO — Financial Figures and Reserves Report as of Jun-30-2026 (PDF, ISAE 3000R report) · Tether/BDO — Financial Figures and Reserves Report as of Mar-31-2026 (PDF, ISAE 3000R report) · Tether — Q2 2026 Press Release (Jul-31-2026) · Tether — Q1 2026 Press Release (May-1-2026) · The Block — Cushion falls by more than 4 billion · CoinDesk — Q2 profit and cushion drop · Decrypt — Q2 results · CoinDesk — Tether hires KPMG (Mar-27-2026) · CoinDesk — Two-year countdown under GENIUS Act · U.S. Treasury — TIC table of major foreign holders · U.S. Treasury — TIC system · Federal Reserve — «Stablecoins in 2025» (Apr-8-2026) · Federal Reserve — «Banks in the Age of Stablecoins» (Dec-17-2025) · Federal Reserve — Historical banking lessons (May-1-2026)