On August 13, Tether announced that more than 650 million people now rely on USDT. Its own Q4 2025 report, published afterward, put year-end users at 534.5 million. The later figure is lower, and it was produced using a deliberately broader counting method. That is not a rounding artifact. It is a methodology problem sitting directly beneath the company's headline claim: that hundreds of millions of people now "own" a decentralized slice of United States government debt.
They do not.
They hold a zero-interest claim on one issuer. That issuer owns the reserves outright, keeps absolute discretion over redemption, and captures every dollar of interest the reserves produce. The token settles. The ownership does not exist in any sense a lawyer or a bankruptcy court would recognize.
I have spent the last four years dissecting reserve structures rather than reading press releases. After Terra, I stopped treating collateralization language as descriptive and started treating it as a claim requiring adversarial verification. This is one of those claims.
Context
Tether is the closest thing crypto has to a central bank, minus the mandate, the deposit insurance, and the transparency. As of June 30, its reported liabilities stood at $183.64 billion, backed by $187.75 billion in reserves — a surplus of $4.11 billion, or roughly 2.24% overcollateralization. Within that reserve, $114.96 billion sits in direct US Treasury bills, with a further $18.63 billion in overnight reverse repos collateralized by approximately $18.60 billion of Treasuries.
The scale is not the story. The structure is.
Ardoino's framing — that USDT represents a decentralized ownership of US debt — arrives during a bull market in which stablecoin legislation is a live political topic and Tether's political positioning is accelerating. The rhetoric is calibrated for that environment. It is also, on close reading, contradicted by Tether's own published materials. That tension is what deserves dissection.
The first failure: attestation is not audit
Tether's reserves are verified by attestation, not by a full GAAP audit. That distinction is not pedantic. An attestation confirms a snapshot of certain balances. It does not test existence, valuation, or internal controls with the depth an audit requires. It does not reconcile against liabilities continuously. It does not require the examiner to issue an opinion on the adequacy of controls over reserve management.
The attestation firm and the precise scope of engagement are not disclosed in the material under review. Without that, the $187.75 billion figure is a counterparty claim, not a verified fact from an independent examiner. In a system where the entire product is a promise to redeem at par, the verification method is the product.
The second failure: the user count contradicts itself
Tether's 2024 methodology identifies on-chain addresses and accounts as a proxy metric and explicitly acknowledges these are upper-bound estimates, because one person can control multiple wallets. That is methodological honesty. It also dismantles the precision of the "650 million" claim by the company's own admission.
Then the internal contradiction. The August 13 announcement claimed over 650 million users in emerging markets. The later Q4 2025 report estimated 534.5 million year-end users using a broader method. A broader method should produce a larger number, not a smaller one. It produced a smaller one. Either the earlier figure was inflated, or the later methodology is inconsistent with the earlier framing. Neither reading strengthens the ownership narrative.
The material further concedes it cannot confirm whether "650 million" refers to distinct individuals, current holders, or redeemable customers. That is the correct caveat, and it is fatal to the claim being sold.
The third failure: nobody owns the reserves but Tether
This is the core of the piece, so I will be precise.
Tether's terms establish that reserve assets are owned and managed by Tether International. Qualified direct customers hold individual contractual redemption rights at par. Secondary-market holders hold only a token they can sell. Neither class shares in the reserve portfolio's income.
The reserves are overwhelmingly interest-bearing Treasuries. At prevailing short-term rates, a $114.96 billion direct T-bill position plus $18.63 billion in reverse repos generates income measured in billions annually. By the terms, that income flows to the issuer. It does not flow to holders. When I simulated the Curve 3Pool in 2020 to test invariant behavior under a 15% depeg, the lesson was that a mechanism's safety depends on who absorbs the loss when the invariant breaks. Here the mechanism is simpler and the asymmetry is starker: holders bear credit risk, the issuer keeps yield.
The user's nominal return is zero. They exchanged a dollar for a dollar-denominated token and loaned the float to Tether for nothing. Framed that way, the "decentralized ownership" claim is not merely inaccurate. It inverts the actual economic relationship. The holder is an unsecured, non-interest-bearing creditor of a private company. That is the exact opposite of ownership.
The fourth failure: redemption is discretionary
If holders truly owned anything, redemption would be a right. In practice it is a privilege conditioned on thresholds and counterparty approval.
Direct redemption carries a $100,000 minimum and a fee of $1,000 or 0.1%, whichever is greater. Verification is subject to Tether's sole discretion. For the emerging-market user cited as the demographic hero of this narrative, direct redemption is functionally unavailable. They must exit through secondary markets, where the issuer's promise is whatever a buyer will pay, not par.
The broader issue: Tether's terms permit delays and suspensions of redemption services across multiple scenarios. That converts the par peg from an enforceable obligation into a conditional one. The condition is a business judgment made by the issuer, not a term the holder can compel. This is the institutional amplifier of credit risk, and it is the single most under-discussed clause in the entire structure.
There is a parallel to KYC theater I have flagged repeatedly in due diligence work: verification requirements are strict enough to impose real cost on compliant users and porous enough that the determination of who qualifies rests entirely with the issuer. Cost lands on the honest. Discretion lands on the issuer.
The fifth failure: bankruptcy seniority is undefined
Tether's materials, per my reading, do not establish a uniform bankruptcy ranking for every secondary-market holder in every jurisdiction. That means the holder's position in a liquidation is not clearly defined. In insolvency, ranking determines recovery. An undefined ranking is not a neutral default; it is a structural disadvantage, because the issuer's general creditors and the issuer's own management of reserve composition will be sorted out by courts before a secondary holder's claim is.
The Bitcoin ETF review I conducted in early 2024 reached a similar conclusion from the opposite direction. Several issuers presented multi-signature cold-storage architectures that were, functionally, custodial arrangements with better branding. The distinction between "decentralized" and "custodial repackaged" is not a slogan. It is determined by who controls keys, who owns assets, and who bears loss. On all three questions, USDT answers: the issuer.
The buffer is thin and partly opaque
The $4.11 billion surplus is 2.24% of liabilities. That is the entire cushion against reserve impairment before par redemption is threatened. The disclosed T-bill and reverse-repo positions account for roughly $133.6 billion of a $187.75 billion reserve, leaving a substantial undisclosed remainder that may include loans and other higher-risk assets. More subtly, the reverse repos are collateralized by roughly $18.60 billion of Treasuries that may overlap conceptually with the $114.96 billion direct T-bill line. If the disclosure basis does not clearly separate them, the market may be overstating direct Treasury exposure. This is not a proven defect. It is an unverified boundary. And in reserve accounting, unverified boundaries are where losses hide.
The contrarian angle: the bulls are not wrong about everything
Here is where the reflexive skeptic overreaches, and I will correct for it.
USDT is not a Ponzi. The reserves are real Treasury instruments, not reflexively rehypothecated tokens of themselves. The income Tether earns is genuine, not the recycling of new depositor funds. That distinction separates USDT from the algorithmic structures that died in 2022, and it deserves stating plainly, because the loudest critics refuse to.
Ardoino's argument about holder dispersion is also not foolish. Hundreds of millions of dispersed holders are, in aggregate, less likely to coordinate a simultaneous exit than a concentrated cohort. My Curve simulation taught me that behavior under stress — not static distribution — determines stability. Ardoino is right that the trigger for a run is harder to reach when no single holder controls enough to start one.
But notice what that argument concedes. It is an argument about the improbability of a run, not the impossibility of one. And it treats holder dispersion as the protection of last resort, which means the issuer is admitting the protection is behavioral, not structural. A fund's safety case should not rest on the assumption that its investors will be lazy.
There is another blind spot the bulls also miss. Emerging-market demand for USDT is real and economically rational — it is a genuinely useful dollarization tool for populations without reliable banking access. That utility exists regardless of the ownership fiction, and dismissing it weakens every legitimate critique. The problem is not that people use USDT. The problem is that they are told they own something they do not.
Takeaway
The metric that matters is not how many people hold USDT. It is what any one of them can enforce. Ownership is an illusion without immutable proof, and a $183.64 billion liability book does not change that arithmetic.
The next time a stablecoin issuer claims that users own something, open the redemption terms and the bankruptcy provisions before you open the announcement. The announcement is marketing. The terms are the contract. Only one of them survives a courtroom.