BNY tested a Treasury settlement after Fedwire's daily close, using two stablecoin issuers whose reserves it already custodies on both sides of the trade.

The sentence names the two issuers and skips the fact that makes the test possible at all: BNY already custodies both of their reserves.

Rated against what the letter and its coverage actually disclose:

One row is a live, if unverified, transaction. The other three are a custody fact already true before this letter and two dated promises.

The transaction never left the bank that already held both issuers' collateral.

• The test never left BNY's own ledger. Both reserve pools already sat inside BNY's custody, so moving money after hours meant updating internal book entries, not routing a trade through the Fed's own rails.
• Fedwire's calendar is the constraint that's still there. The test worked around Fedwire's daily close; it didn't reopen Fedwire. Any leg that needs a bank outside BNY's custody still waits for business hours.

The gap this test targets is real and structural, not cosmetic. RLUSD and USDO both trade around the clock, but the short-dated Treasuries backing them settle through Fedwire, a system that runs on business days. A redemption wave that lands on a Saturday hits collateral that is, in the strictest sense, frozen until Fedwire reopens. BNY's test is a bet that the fix sits at the custodian layer, not at the Fed.

BNY didn't need the market's cooperation to run this test, because BNY already is the market for these two issuers. RLUSD's reserves and OpenEden's Treasury fund both sit in BNY's custody today, which is what let the bank move assets after Fedwire's close without a second institution's rails involved at all.

That framing cuts both ways. It's a genuine capability, proven once, inside one bank's books. It says nothing yet about whether a stablecoin issuer whose reserves sit with a different custodian, or split across several, gets the same after-hours flexibility, since that would require either Fedwire itself to extend its calendar or every custodian in the chain to offer the same internal workaround BNY just demonstrated on its own clients.

• No transaction size or date disclosed. The client letter describes the mechanism, not the trade. There is no dollar figure, timestamp, or confirmation independent of BNY's own account.
• Both issuers already bank with BNY. RLUSD and USDO aren't a representative sample; they're BNY's own custody clients. Whether a custodian without an existing relationship on both sides of a trade can replicate this is untested.
• Fedwire itself didn't move. The test avoided Fedwire's daily close; it didn't extend it. Any leg of a transaction that needs a bank outside BNY's custody still waits for Fedwire's business hours.
• "Always-on" is a 2027 promise with a 2026 placeholder in between. Tokenized Treasury pilot trades are dated only "by the end of 2026," with no named counterparty or asset size attached yet.

BNY's reach into stablecoin reserves is not new. Tracee's 11 July briefing noted that USDC's reserve management stays with BNY Mellon even after Circle's own OCC-chartered trust bank opened. Tracee's coverage of the five-custodian reserve-fund race counted BNY Dreyfus's BSRXX fund alongside BlackRock, Goldman Sachs, JPMorgan, and State Street as one of five structurally identical products competing for the same issuer cash. RLUSD and USDO now add a second data point: BNY isn't just one custodian among several, it's the custodian sitting on both sides of a transaction between two issuers who compete with each other for the same dollar-stablecoin market.

The 24/7 ambition also has a mirror image already in the record. Tracee's 10 July briefing on SWIFT's blockchain ledger found the same asymmetry from the other direction: tokenized deposits already move continuously on that ledger, but final settlement still clears through SWIFT's correspondent banking rails underneath it. BNY's test is the same seam approached from the Treasury side: the token or reserve leg wants to run non-stop, and the settlement leg it ultimately depends on, Fedwire in this case, still runs on a five-day week.

As stablecoin reserves scale, a redemption promise that quietly depends on the day of the week is a liquidity question regulators, not marketers, will end up counting the hours on.

BNY's test proves a large custodian can move Treasury-backed reserves on its own ledger outside Fedwire's operating calendar, when both sides of the trade already bank with it. It does not prove Fedwire's calendar is changing, and it does not prove a stablecoin issuer whose reserves sit elsewhere gets the same flexibility. Treat this as a real capability demonstrated once, on the narrowest possible sample, with a 2027 target attached that depends on scaling far beyond it.

Watch three things:

• Whether BNY discloses a transaction size, date, or a counterparty bank outside its own custody book for this or a follow-up test.
• Whether BNY's private-blockchain tokenized Treasury pilot ships by its stated end-of-2026 deadline, and with which counterparty.
• Whether the Federal Reserve extends Fedwire Securities Service's own operating calendar, the one dependency this test worked around rather than removed.