ANALYSIS – This piece lays out an analyst thesis on where wholesale digital-ledger projects could be pushing the banking system, not a confirmed plan. The underlying facts about Project Agorá, mBridge, and current interbank rates are sourced and current as of publication; the “endgame” argument is Bankinfobook’s own reading of the plumbing, not an official policy disclosed by any central bank.

On May 27, 2026, the Bank for International Settlements published something that read, on its surface, like a routine technical milestone: a prototype report confirming that tokenized central bank reserves and tokenized commercial bank deposits can settle across borders, atomically, in seconds. Seven central banks the New York Fed, Bank of England, Bank of Japan, Banque de France, Swiss National Bank, Bank of Korea and Bank of Mexico signed off, alongside more than 40 private institutions including JPMorgan, HSBC, Deutsche Bank, Mastercard and SWIFT itself. The Bank of Canada joined shortly after.

Read as an engineering milestone, it’s unremarkable: faster settlement, fewer intermediaries, less reconciliation risk. Read as plumbing, it looks like something else, a redesign that could quietly reshape which banks survive the next credit cycle.

The mechanism: atomic settlement’s hidden cost

Project Agorá’s core innovation is atomic settlement: a cross-border payment either completes in full, instantly, across every leg of the transaction, or it reverses entirely. There’s no partial fill, no overnight float, no multi-day correspondent chain.

That’s the selling point. It’s also the catch. Atomic, all-or-nothing settlement requires the full value of a transaction to be pre-funded and locked before it moves, there’s no daylight overdraft, no sequential netting, no credit-line float to lean on while cash arrives. The legacy correspondent banking system runs on exactly that kind of float. Mid-tier and regional banks depend on it to manage multi-currency exposure without parking enormous locked balances across jurisdictions they don’t have the balance sheet to support.

The BIS report itself flags unresolved questions in this area liquidity-saving mechanisms, cybersecurity, and settlement-finality governance are explicitly listed as unfinished work, not solved problems. In plain terms: nobody has fully worked out how a mid-sized bank is supposed to fund instant, all-or-nothing settlement without tying up capital it currently puts to work elsewhere.

The BIS has an answer, but it comes with a catch

To be fair to the architects: the BIS is not blind to the pre-funding problem. Liquidity-Saving Mechanisms (LSMs) queue-based netting and gridlock-resolution algorithms that let payments offset against each other rather than requiring every participant to post full collateral upfront are an established, decades-old tool in real-time gross settlement systems, and BIS-convened discussions on Agorá have explicitly flagged LSMs as part of the design conversation for reducing the liquidity burden of atomic settlement.

Here’s the catch the BIS’s own research is candid about: LSMs work by letting non-urgent payments wait in a queue to be netted against offsetting flows. That’s the mechanism’s entire value proposition and it’s also its limit. Netting takes time. The more a system leans on LSMs to conserve liquidity, the more it reintroduces exactly the kind of settlement delay that atomic, “instant” settlement was supposed to eliminate. A payment can be cheap and queued, or instant and fully pre-funded, pushing hard on one side of that trade-off pulls against the other.

That tension is where a further-out possibility opens up, worth stating plainly as inference rather than fact: someone still has to float the liquidity for payments that need to clear instantly rather than wait in a netting queue. The institutions best positioned to do that at scale are the same handful already sitting inside the Agorá consortium with the balance sheets to support it JPMorgan, BNY Mellon, HSBC, Citi. Nothing in the public Agorá materials describes this as a designed feature, but the economics point toward it: a mid-tier bank facing an urgent cross-border payment and an underdeveloped netting queue has an obvious workaround pay a Tier-1 institution a fee or spread to pre-fund the instant leg on its behalf, effectively “renting” atomic-settlement liquidity rather than holding it.

If that’s how it plays out, the mid-tier bank doesn’t get frozen out of Agorá outright. It gets absorbed as a fee-paying dependent of the same institutions it used to compete with on correspondent banking a softer, slower version of the squeeze, but a squeeze all the same.

Note: the “liquidity-as-a-service” role described above is Bankinfobook’s inference from the LSM/atomic-settlement trade-off  no central bank or Tier-1 institution has stated this as an intended design.

The rates make the “rental” option look worse than it sounds

That liquidity-as-a-service arrangement isn’t happening in a vacuum it’s landing on top of a mid-tier funding squeeze that’s already visible in the data. Three-month Euribor is sitting around 2.35% as of early July 2026, roughly a percentage point above where the ECB’s own official rate anchors, and up sharply from a year earlier. Twelve-month Euribor is running even higher, near 2.7%. That gap between what banks pay each other overnight and what the central bank charges is a direct read on interbank funding stress and it falls hardest on exactly the institutions least able to absorb it: the regional and mid-tier banks without the deposit base or capital markets access of a JPMorgan or a BNY Mellon.

Paying a Tier-1 institution a fee to pre-fund atomic settlement is a manageable cost in isolation. Paying it on top of already-elevated interbank funding costs is a different proposition it’s a second toll layered on a bank whose margins are already compressed. That’s the setup in which corporate treasury clients, watching their bank’s transaction pricing climb, start looking at the private rails instead.

Where the deposits actually go, and it isn’t retail stablecoins

The “private rails” half of this thesis needs a correction, and it’s worth making explicitly: for a genuine $100 million cross-border corporate transfer, retail stablecoins like USDC or USDT are not the credible substitute they’re sometimes made out to be. Regulatory clarity has improved sharply in 2026, the GENIUS Act in the US and MiCA in the EU have narrowed the uncertainty that used to keep institutional treasurers away and stablecoin B2B volume has genuinely grown, up an estimated 733% year-over-year according to McKinsey/Artemis data. But that growth concentrates in corridors where legacy banking is worst Latin America, parts of Africa and Southeast Asia, Turkey-UAE not in the well-banked corridors where Agorá and atomic settlement are aimed. A large multinational moving money between London and New York isn’t reaching for USDC.

The more accurate version of the bypass mechanism is already under construction, and it doesn’t route through crypto rails at all. In June 2026, JPMorgan, Citi, Bank of America and Wells Fargo confirmed they’re building a shared Tokenized Deposit Network (TDN) through The Clearing House, targeting a first-half 2027 launch. The pitch is explicitly defensive: keep large corporate deposits inside the regulated banking system, on a blockchain-speed settlement layer, so there’s no opening left for either a government CBDC or a stablecoin issuer to capture that flow instead. JPMorgan’s Kinexys platform (formerly JPM Coin) already moves several billion dollars daily for institutional clients; Citi’s Token Services runs live cross-border transfers between New York, London and Hong Kong; BNY launched its own tokenized deposit service in January. Citi’s Institute has projected tokenized bank deposits could support $100-140 trillion in annual flows by 2030.

The detail that matters most for this thesis: the TDN’s founding members are the same handful of Tier-1 institutions already sitting inside Project Agorá. Regional and mid-tier banks are not named participants. A Bank of America executive was candid enough to note that clients aren’t yet “beating down the door” for tokenized deposits this is infrastructure being built ahead of demand, positioning the biggest banks to capture large corporate treasury relationships before smaller competitors have any comparable rail to offer.

That’s the same institutions, twice over. If a mid-tier bank ends up renting instant-settlement liquidity from JPMorgan or Citi under Agorá, and its largest corporate clients are simultaneously migrating their treasury deposits onto JPMorgan’s or Citi’s own tokenized network, the regional bank isn’t losing ground to some external “shadow grid.” It’s losing ground to the same small set of megabanks on two fronts at once as a liquidity landlord and as a deposit destination.

Note: the “two fronts” framing connecting Agorá and the Tokenized Deposit Network is Bankinfobook’s own reading of the overlap between these two separate, independently confirmed developments the banks involved have not described them as a coordinated strategy.

Where this points, and where it doesn’t

None of this means regulators or megabanks are engineering mid-tier bank failures. The BIS report is explicit that Agorá is designed to preserve the two-tier banking system and the “singleness of money” not replace commercial banks. The project is wholesale-only, not retail, and central banks have been vocal about keeping domestic monetary control rather than ceding it to a shared ledger. The Tokenized Deposit Network’s own backers describe it in defensive terms too protecting deposits from stablecoins and a hypothetical CBDC, not from other banks.

But intent and outcome aren’t the same thing. A system that makes atomic, pre-funded settlement the new baseline for cross-border wholesale payments, paired with a liquidity-rental role only the largest banks can fill, paired with a tokenized deposit rail built by and so far limited to those same largest banks, adds up to a set of overlapping advantages that don’t require any coordinated intent to produce a consolidating effect. Regional and mid-tier banks aren’t being frozen out by decree. They’re facing a landscape where the fastest, cheapest, most liquid rails for large corporate clients increasingly belong to a small number of institutions that also happen to be their competitors.

The prototype phase is done. Real-value testing on Agorá is next, with no confirmed production timeline. The Tokenized Deposit Network targets first-half 2027. The EU’s Pontes framework, aligned with Agorá’s design, is slated to connect into Europe’s TARGET payment infrastructure later this year. Whether mid-tier banks get meaningful access to these rails on comparable terms or whether they remain locked out while the largest institutions build the toll roads is the open question worth watching, not a settled outcome.