Paxos: Regulated Issuance and Settlement Infrastructure

As of November 1, 2026, DefiLlama put Paxos-issued supply at roughly $5.7 billion across three dollar tokens, split as about $3 billion of PayPal USD, about $2.9 billion of Global Dollar, and roughly $22 million of Pax Dollar, Paxos's own brand. That last number is the architecture. The own-brand float is a rounding error and the business is issuing other companies' money, which makes Paxos the clearest available example of regulated issuance exposed as a service.

Four Parties, One Token

Take PYUSD and list the parties a holder is actually dealing with. PayPal owns the brand and the customer relationship. Paxos Trust Company, N.A. issues the token, holds the reserves and bears the redemption obligation. The Office of the Comptroller of the Currency supervises the issuer, following Paxos's conversion from its 2015 NYDFS limited purpose trust charter to a national trust charter on December 12, 2025. Reserve custodians hold the assets, and a separate accounting firm attests to them monthly.

None of those is PayPal, and the obligation is the part that gets misread. When CNBC covered a PYUSD incident it had to append a clarification stating that while PayPal agrees to redeem PYUSD, Paxos is the party obligated to redeem one-for-one. A financial news desk got the obligor wrong, which is a reasonable signal that most integrators will too.

The same pattern repeats across jurisdictions with different issuers behind one ticker. Global Dollar is issued by Paxos Digital Singapore as a Major Payment Institution supervised by the Monetary Authority of Singapore, and by Paxos Issuance Europe under Finland's FIN-FSA in compliance with MiCA, with a separate ADGM-licensed entity covering the UAE. The ticker is constant. The obligor and the supervisor change with geography.

The Primitives and Who Holds Them

Strip the product away and the platform exposes four primitives: mint, burn, freeze, and attest. Everything else is distribution.

Mint and burn are the issuance pair, and they are privileged operations controlled by the issuer's keys rather than by any holder or by the brand. Freeze is an administrative capability that exists in the contract and sits with the issuer. Attest is the periodic third-party confirmation that reserves match outstanding supply, performed monthly.

The trust assumptions follow directly from that list and they are not evenly distributed. A holder trusts the issuer's solvency and the fiduciary character of a national trust charter, the reserve custodians, the attesting firm, and the issuer's operational control over the mint key. The brand is trusted for none of it, and the holder's recourse does not run to the brand at all.

Value Flows the Opposite Way to the Obligation

Reserve interest accrues to the issuer, which is where the revenue in this model originates. The issuer then shares revenue with the brand, which owns the distribution. The holder receives no interest from the issuer, because the GENIUS Act prohibits it.

So the economics flow from issuer to brand while the obligation flows from holder to issuer, in opposite directions along the same chain. That is the structural fact worth carrying away, and the regulator has now drawn a line inside it. In a 376-page rulemaking proposal published in late February 2026, the OCC expanded the statutory interest ban to reach certain indirect pathways while explicitly permitting two things: independent payments by merchants, and issuer payments that share revenue with a third party for white-label issuance.

Read together, that preserves the structure and cuts its last link. Revenue sharing between the issuer and the brand survives. The arrangement in which the brand pays rewards to holders does not, even though the issuer never paid interest directly. The white-label model is intact; the yield pass-through to the end holder is the part being closed.

The Mint Is the Blast Radius

On October 15, 2025, at 3:12 p.m. Eastern, Paxos minted 300 trillion PYUSD during an internal transfer and burned it 22 minutes later, describing it as an internal technical error with no security breach. The two preceding transactions had each been for 300 million, which is what made a keying error the obvious reading.

The architectural lesson is about bounds rather than embarrassment. Mint authority had no effective ceiling, so a single operation could create a notional larger than global output, and the control that caught it was a human noticing rather than a limit refusing. Downstream systems had to defend themselves: a lending protocol froze new deposits of the token while it assessed exposure. No holder lost anything.

Two readings are both fair. One is that concentrated mint authority in a centralized issuer is a systemic exposure, and that internal controls permitting an unbounded mint were inadequate. The other, argued at the time, is that the error was visible to anyone with a block explorer within seconds and corrected in under half an hour, which no conventional ledger would have allowed. The honest conclusion is that public issuance converts an operational control weakness into a public, fast-correcting one rather than eliminating it, and a rate or magnitude limit on the mint path remains the mitigation that should have existed.

Metrics That Describe the Business, and What Is Improving

Supply is the wrong headline metric here. The informative ones are the split between own-brand and partner-issued float, which at roughly $22 million against about $5.9 billion says the company is an issuance platform rather than a token sponsor, and the count of distinct regulated issuing entities, which is what determines how many jurisdictions a partner can launch into.

The threshold worth tracking is $10 billion. Under GENIUS, state supervision is limited to issuers at or below that level of consolidated outstanding issuance and larger issuers move to the federal framework, with the OCC regulating national banks, foreign issuers and any stablecoin above the line. Paxos sits around $5.7 billion and already converted federally, so it crossed that supervisory boundary by choice rather than by growth, which is the more interesting data point.

What is improving is that the obligor is now identifiable and federally supervised, the attestation cadence is monthly, and the regulator is writing explicit rules about which payments in a white-label chain are permitted. For a technical owner integrating any of this, the diligence target is the issuing entity for the specific jurisdiction, not the brand on the token, and the question to ask about the mint path is whether a magnitude limit now refuses what a person previously had to catch.

For informational purposes only. Not an offer to buy or sell any security. Available only to accredited investors who meet regulatory requirements.

‍