Nearly two-thirds of banks are building corporate deposit tokens as their answer to stablecoins

A deposit token is worth the number of counterparties willing to receive it, and on that test the two architectures in this map are not comparable instruments. Participate brings 600 member banks and a flow that already exists, loan participations, to a token Custodia and Vantage have had live since last March ^[10][12]; Cari brings five named design partners and the stated goal of getting banks onto an interoperable network ^[14], which works out to 120 member banks for every design partner ^[21]. One is distribution bolted to volume that has to settle anyway. The other is standard-setting, which is a build.

The survey figure is a share of institutions, not a share of payments ^[1], and the leftover is the part that prices the asset: about one bank in three is neither offering nor developing one ^[19], so a corporate treasurer moving value across bank boundaries keeps hitting an endpoint that has to be paid the old way. N3XT's NDD is engineered for that gap, letting non-banks hold and transact without opening an N3XT account and deliberately breaking what Scott Shay calls the walled garden of bank tokenized deposits ^[6] - the shape of program run inside a bank for its own clients, of which JPMorgan's Kinexys is the largest example named here ^[13]. The backing is cash or very short duration Treasuries matched one-for-one, with regulatory capital held on top ^[4], and settlement is a few seconds ^[5].

Citi is the useful tell, because it is running both sides: selling 24/7 clearing and tokenized deposits to Siam Commercial Bank ^[17] while exploring reserve management, stablecoin conversion and a Citi stablecoin of its own ^[16]. That is a bank declining to pick, which is what you do when you cannot price the outcome yet.

This is probably wrong, but my read is that the two-thirds number is option-buying denominated in engineering hours rather than a claim on cross-border share, and the programs that convert will be the ones attached to a flow rather than the ones with the cleanest standard. Or rather, the more interesting version: the winning distribution may not be a consortium at all but one large bank's rails, and Citi signing a second and third user after Siam Commercial ^[17] is the evidence that would show it. Two other paths are open. If interoperability actually arrives, the thing Capgemini's Michael Levens says attention is turning toward alongside liquidity management and commercial models ^[18], the walled garden stops mattering and the survey number becomes a real network. If it does not, these stay internal treasury tools and the cross-border payment goes to whatever the counterparty already accepts, which is Shay's warning about banks that stay out ^[7]. What would break the thesis is any of Cari's five design partners settling third-party volume without a pre-attached flow ^[14]. Until then everything here is valued on architecture, because the map reports no volume or balance figures for any of it ^[22].