The DvP logic of Part 2 has a direct FX analogue: Payment versus Payment, or PvP. The asset leg becomes one currency; the cash leg becomes the other; and the same atomicity property either holds or does not. The historical accident that made PvP an institutional priority is the failure of one mid-sized German bank in June 1974, and the institutional response (CLS Bank, founded 28 years later in 2002) is now the entity that tokenised FX rails are quietly positioned to compete with.
Herstatt risk and the 1974 origin
The textbook FX settlement risk is named for Bankhaus Herstatt, a private German bank ordered into liquidation by the Bundesaufsichtsamt für das Kreditwesen on 26 June 1974, at 15:30 Frankfurt time. The bank's counterparties had paid Deutschmarks earlier in the day on the assumption that they would receive US dollars in New York later, after New York opened. They had performed; Herstatt had not. When Herstatt was wound up, the dollar leg never settled. Counterparties lost the open exposure on every trade where they had paid but not yet received.
The losses were not enormous in absolute terms (a few hundred million dollars), but the structural lesson was. The settlement of an FX trade had a window, and inside it one party could fail. The window was not days; it was hours, defined by the time-zone gap between the two currency-clearing systems. Every FX desk worldwide carried the equivalent risk on every open trade.
The post-Herstatt institutional response took nearly three decades to arrive. CLS Bank (Continuous Linked Settlement), a New York-chartered Edge Act bank owned by a consortium of major FX-active institutions, began operations in September 2002 specifically to provide PvP for major-currency FX trades. Each side of the trade pays funds into CLS's accounts at the relevant central banks, and CLS releases the corresponding amount only when the matching pay-in has been confirmed. The Herstatt window closes because no party releases until the matching release is guaranteed.
CLS today settles 18 currencies and a substantial majority of the world's eligible spot, forward, and FX swap volume. It is one of the systemically important financial market infrastructures in the international monetary system, supervised by the Federal Reserve and overseen by an international cooperative oversight arrangement.
Tokenised PvP at the smart-contract level
The tokenised FX architecture does to PvP what tokenised DvP does to securities settlement: replicate the all-or-nothing property at the protocol level, in seconds, on demand, transaction by transaction. Two tokenised currencies sit on the same ledger (or are linked across ledgers via a coordinator pattern). The smart contract that debits one balance and credits the other is a single indivisible operation. Either both legs commit, or both revert. The Herstatt window does not exist because there is no interval in which one currency has been paid and the other has not.
Kinexys Digital Payments runs production on-chain FX between supported currency pairs, settled atomically in commercial bank money rather than CLS-routed central bank funds. The November 2024 rebrand explicitly added on-chain FX between USD and EUR as a headline product line. The mechanics are the same as for the cash leg of any tokenised DvP transaction: the protocol enforces simultaneity; the off-chain treasury operations of JPMorgan back the on-chain balances.
The Asia case study most often cited is the Project Guardian SGD/JPY DeFi pilot, in which JPMorgan's Onyx (now Kinexys) collaborated with DBS and SBI to execute live FX trades and tokenised JGB transactions atop a permissioned variant of an Ethereum DeFi protocol. The trial demonstrated that the PvP atomicity property could be inherited from public-chain DEX semantics into a permissioned setting acceptable to bank-grade counterparties, and it remains one of the most cited proofs of concept in the Project Guardian catalogue. Project Agorá, the BIS-led wholesale cross-border initiative, generalises the same principle: tokenised commercial bank money on a unified ledger, settling cross-currency at the protocol layer rather than through a network of correspondent banks.
The strategic CLS-cannibalisation tension
CLS is a member-owned utility, and its owners are the same institutions building tokenised FX rails. JPMorgan, HSBC, Citi, BNY, Goldman, Morgan Stanley, Deutsche Bank, the Japanese megabanks, the Singapore banks: every major CLS member is also a participant in at least one tokenised FX programme. The same balance sheets fund both layers.
If tokenised FX matures along the trajectory Project Guardian and Project Agorá are pushing, it competes with CLS at the wholesale FX settlement layer. The economics are favourable: tokenised PvP avoids CLS membership fees and pay-in funding, settles in seconds rather than waiting for the New York close, and does not require pre-funding the CLS account in each currency. For the issuing banks, tokenised FX captures revenue that previously bypassed them through the CLS utility model.
The tension is not yet acute. Tokenised FX volumes are a small fraction of CLS's daily $7-8 trillion. But the trajectory is structural. The same banks that fund CLS are building the rails that, on a long horizon, route around it. Whether CLS evolves (it could itself become a tokenised-rail operator and capture the new flow) or is gradually disintermediated is the strategic question, and it is being worked out one currency pair at a time. The practical vocabulary is the same as for tokenised DvP: this is not a new settlement primitive, it is the same post-Herstatt PvP property implemented on infrastructure that did not exist in 2002, compressing the funding cost of atomic settlement to a level the historical response could not match.
Part 4 fixes the precise nomenclature that operators need to keep all of this clear: T+N, intraday, near-real-time, and atomic, and why marketing language conflates them.