The End of Settlement Risk? Inside the Promise of Atomic Payments

Atomic settlement can remove the principal risk created when one leg of a trade completes before the other. But it cannot eliminate liquidity, legal, operational or systemic risk—and the race to deploy it at scale is only beginning.

The most dangerous moment in many financial transactions is not when a price is agreed. It is the interval between one side delivering and the other side doing the same. A bank can send one currency and wait for another. An investor can release cash while a security is still moving through a separate system. A tokenised asset can change ownership on one ledger while the money needed to pay for it sits on another. The gap may last seconds, hours or a full settlement cycle, but the economic problem is the same: one party can perform while the other fails.

That is settlement risk, and finance has spent decades trying to compress it. The failure of Bankhaus Herstatt in 1974 made the foreign-exchange version famous because counterparties had paid Deutsche marks but did not receive the dollars due later in the day. Half a century later, the problem is smaller relative to the scale of markets but far from gone. BIS analysis of the April 2025 Triennial Survey found that more than $14 trillion of gross FX obligations were settled on an average day; $1.4 trillion, or 10%, still settled gross bilaterally without a mechanism that eliminated or mitigated FX settlement risk.

Atomic settlement promises a more radical answer. Instead of making one payment and trusting that the second leg will arrive, the transaction is designed so that all linked transfers happen or none of them do. In foreign exchange that is payment versus payment (PvP). In securities it is delivery versus payment (DvP). In tokenised markets, smart contracts and synchronisation mechanisms can extend the same logic across several assets, currencies and ledgers. The phrase sounds like the end of settlement risk. The reality is more interesting: atomicity can eliminate one of the oldest forms of principal risk while shifting other risks into liquidity, code, interoperability and legal design.

What Atomic Settlement Actually Does

Atomic does not necessarily mean “instant”, and it does not require a blockchain. It means conditional and indivisible execution. The cash leg settles if and only if the asset or second payment leg settles. If a required condition fails, the exchange is not completed. The CPMI’s work on payment versus payment treats this structure as the strongest way to remove FX principal risk because the final transfer of one currency is linked to the final transfer of the other.

The distinction matters because speed alone can be deceptive. Two payments can each settle in milliseconds and still expose a participant if they are independent. Conversely, two systems can be technologically different and still settle atomically if a trusted mechanism coordinates them. Project Meridian FX demonstrated that point in 2025: a technology-neutral synchronisation operator coordinated PvP settlement between different wholesale payment infrastructures, including links between an emulated UK RTGS system and experimental Eurosystem solutions.

The economic benefit is straightforward. A buyer should not lose principal because it paid and the seller failed to deliver; a seller should not deliver an asset and then discover the payment never arrived. But the protection is narrow. The BIS emphasises that PvP does not eliminate replacement-cost risk—the cost of replacing a failed trade at a new market price—or liquidity risk when expected funds do not arrive on time. Atomicity changes the failure mode; it does not abolish every consequence of a counterparty default.

What atomic settlement reduces—and what it leaves behind

The FX Market Shows Why the Problem Is Not Solved

Foreign exchange remains the clearest laboratory because the transaction inherently contains two payment legs. The BIS’s June 2026 stocktake of FX settlement risk found that 36% of average daily settlement in April 2025—about $5.2 trillion—used PvP and therefore eliminated FX settlement risk. Another 54% used intragroup settlement, pre-settlement netting or timing controls that mitigate rather than eliminate it. The remaining 10%, more than $1.4 trillion per day, settled gross bilaterally and was exposed to the full principal amount.

That is a striking result for two reasons. First, the residual risk is enormous in absolute terms even after decades of public- and private-sector effort. Second, the problem is not simply lack of technology. A quarter of the gross bilateral amount was eligible for PvP but still did not use it. Other trades were excluded because counterparties lacked access, currency pairs were unsupported or trade types fell outside existing arrangements.

The survey also highlights a new timing problem. Some PvP systems are built around cut-off times and next-day settlement, which makes same-day trades difficult to accommodate. As securities markets shorten settlement cycles, institutions can need same-day FX swaps to fund cross-border securities purchases. Faster securities settlement can therefore create greater pressure for faster and more flexible FX settlement. Atomicity has to become more available, not merely faster where it already exists.

From Specialist PvP to Programmable Atomicity

Traditional PvP and DvP arrangements already prove that atomic settlement is not a new concept. What tokenisation changes is the range of transactions that can be coordinated and the amount of workflow logic that can be embedded around settlement. The IMF’s 2026 note on Tokenized Finance describes programmable shared ledgers as capable of collapsing trading, compliance, settlement and reconciliation into synchronised processes. That matters because many settlement frictions originate not in the final book entry but in the sequential checks and messages that precede it.

The BIS Annual Economic Report 2026 similarly argues that tokenisation can align instruction, compliance checks and settlement in coordinated workflows. In cross-border payments, the attraction is not simply that a token moves faster than a correspondent-banking message. It is that the necessary balances can be reserved, conditions verified and multiple payment legs released in an all-or-nothing transaction.

This is where the term “atomic payments” becomes broader than a crypto-market slogan. It can describe a shared ledger, but it can also describe synchronisation between existing RTGS systems and new asset ledgers. The strategic question for banks and market infrastructures is therefore not blockchain versus legacy technology. It is which architecture can deliver conditional settlement with legal finality, resilient access to safe money and enough interoperability to work across real markets.

Project Agorá Moves the Concept Toward Real-Value Testing

The most important 2026 proof point is Project Agorá, the BIS-IIF public-private experiment on wholesale cross-border payments. Its prototype combined tokenised commercial-bank deposits with tokenised central-bank reserves and demonstrated atomic multi-currency settlement across jurisdictions. The project involved seven central banks and more than 40 regulated financial institutions during the prototype phase.

Agorá is significant because it preserves the two-tier monetary structure rather than replacing banks with a single supranational settlement token. Commercial banks continue to hold client-facing liabilities; central-bank reserves remain the settlement anchor. Smart contracts coordinate the workflow. The BIS says the design could support around-the-clock wholesale cross-border payments if implemented, while embedding conditional triggers and compliance logic directly into transactions.

The project is not production infrastructure yet. In May 2026, the BIS said the next stage would advance to real-value testing for selected currencies and participants, with the Bank of Canada joining. That distinction between prototype and live settlement is critical. Demonstrating atomicity in a controlled environment is a technical milestone; scaling it requires liquidity arrangements, governance, operating rules, cyber resilience and legal recognition across multiple jurisdictions.

Atomic-settlement initiatives: where the major projects stand

Existing RTGS Systems Are Learning to Synchronise

Europe and the United Kingdom illustrate a pragmatic path: keep central-bank RTGS infrastructure and make it interoperable with tokenised markets. The ECB’s Pontes programme is designed to link market DLT platforms with TARGET Services. Its architecture provides a route for DvP and other all-or-none transactions while keeping final settlement of the cash leg in central bank money. The ECB continues to target an initial launch in the third quarter of 2026.

That timetable is now operational rather than theoretical. At a 22 July 2026 ECB focus session on Pontes, the Eurosystem said Pontes functionality was being tested internally and user testing was scheduled to start in August. The immediate aim is not to replace TARGET with one giant shared ledger. It is to bridge DLT-based market infrastructures to the settlement asset institutions already trust.

The Bank of England is pursuing a related synchronisation model for its renewed RTGS service, RT2. In May 2026 it launched a Synchronisation Lab with 18 organisations exploring use cases including PvP for cross-border spot FX, DvP for tokenised securities, collateral optimisation and house purchases. The Bank’s June 2026 work on extending RTGS and CHAPS settlement hours makes the link explicit: longer operating hours are a necessary complement if synchronised settlement is to support tokenised markets and FX use cases that may need central-bank money outside today’s operating windows.

Australia Shows There Is More Than One Form of Settlement Money

Atomic settlement also raises a harder question: what money should sit on the cash leg? Central-bank money is the safest settlement asset, but not every market will move entirely onto a wholesale CBDC. Australia’s Project Acacia deliberately tested several possibilities across 20 wholesale tokenised-asset use cases: existing RBA exchange settlement account balances, a pilot wholesale CBDC, tokenised commercial-bank deposits and stablecoins.

The finding is important because it suggests atomicity is a settlement property rather than a single-money model. Different assets and market structures may use different settlement instruments, provided interoperability, legal rights and risk controls are clear. Acacia found potential efficiency and counterparty-risk benefits, but it also identified scaling challenges that require continued work by regulators and industry.

That plural model is likely to matter globally. A tokenised bond may settle in central-bank money through an RTGS bridge; an institutional tokenised-deposit transfer may settle across a permissioned bank network; a low-value cross-border transaction may use a different PvP design. Project Rialto tested yet another configuration, combining instant PvP, automated FX and tokenised central-bank money in a simulated cross-border payment infrastructure focused on low-value payments.

The Liquidity Paradox: Less Credit Risk, More Real-Time Funding Pressure

Atomic settlement can make a trade safer while making liquidity management more demanding. Traditional settlement cycles create exposures, but they also create time. Institutions can net obligations, mobilise collateral, borrow cash and sequence payments before final settlement. Remove the lag and the balance has to be ready when the transaction is ready.

The IMF’s 2026 analysis describes this as a redistribution of risk. Atomic DvP reduces counterparty credit risk and operational frictions, but the loss of settlement lags and end-of-day netting shifts liquidity demand into continuous real time. Credit exposure falls; intraday liquidity needs can rise. In a 24/7 environment, liquidity backstops designed around business-day cycles may also need to evolve.

This is one reason “instant everything” is not automatically optimal. Gross atomic settlement can consume more liquidity than systems that net many transactions before settlement. Modern designs therefore need liquidity-saving mechanisms, queue management and intelligent reservation of balances. Meridian FX explored liquidity-saving functionality alongside atomic PvP, while Agorá locks the balances required before releasing the transaction. The problem moves from “Will my counterparty deliver?” toward “Can I mobilise the right asset, in the right currency, on the right ledger, at the exact moment the transaction becomes executable?”

Atomicity Does Not Create Legal Finality

A ledger can declare a transaction complete before a court is prepared to treat it as legally final. That gap is especially important across borders. Property law, insolvency law, settlement-finality rules and the legal status of tokenised records still differ by jurisdiction. If one side of a transaction becomes insolvent during a dispute, technical irreversibility is not enough; participants need to know which rights survive and which record is legally authoritative.

The IMF identifies legal certainty as a prerequisite for scaling tokenised finance, including clarity over whether tokenised records constitute definitive ownership and whether ledger finality is recognised in law. The CPMI PvP report reaches a similar conclusion from the FX side: differences in national legal and regulatory frameworks remain a barrier to broader PvP adoption.

Pontes is designed with this issue in mind. The ECB states that the final cash settlement occurs in T2, providing the legal certainty associated with existing TARGET infrastructure while the DLT leg is synchronised through an interoperability mechanism. That hybrid approach may prove more important than debates over whether the future infrastructure is “on-chain” or “off-chain”. Institutions need a settlement outcome that is final in code, in operations and in law.

The New Risks Live in Code, Connectivity and Governance

Atomicity can reduce hand-offs, but it can also increase the blast radius of a technical failure. If multiple legs, compliance conditions and collateral actions are linked through code, an incorrect rule or faulty data input can affect the entire transaction at machine speed. The IMF warns that the same characteristics that make smart contracts efficient—speed, determinism and automation—can amplify coding errors and bad data before humans have time to intervene.

Interoperability is another risk frontier. A future market is unlikely to run on one universal ledger. Public networks, bank-led permissioned platforms, central-bank systems and legacy infrastructures will coexist. The BIS 2026 monetary-system analysis notes that fragmented networks can force reliance on ad hoc bridges and create operational vulnerabilities. An atomic trade is only as robust as the connections that make every leg part of the same conditional outcome.

Governance therefore becomes part of settlement engineering. Who can pause a smart contract? Who changes the code after a vulnerability is discovered? Who decides whether a failed oracle reading should trigger or cancel settlement? Who has access to central-bank money at 3 a.m.? Those questions are not implementation details. They determine how a system behaves under stress, which is when settlement infrastructure matters most.

What Atomic Payments Mean for Banks, Fintechs, Regulators and Investors

For banks, atomic settlement can reduce counterparty exposures, reconciliation costs and trapped collateral, but it raises the strategic value of real-time treasury capability. A bank that can execute tokenised transactions but cannot source liquidity across currencies and platforms continuously will not capture the full benefit. Treasury, payments, collateral management and technology functions are likely to converge more closely as settlement becomes programmable.

For fintechs and infrastructure providers, the opportunity is less about inventing another token and more about orchestration. Meridian FX shows that synchronisation can be technology-neutral; Pontes is explicitly an interoperability bridge. That creates room for firms specialising in connectors, compliance logic, liquidity routing, identity, ledger integration and synchronisation services. But systemically important orchestration will attract systemically important expectations around resilience, governance and supervision.

For regulators and central banks, the objective is not to mandate atomicity everywhere. It is to expand the cases where principal risk can be removed without creating worse liquidity or operational outcomes. The BIS FX data show the immediate policy opportunity: broaden access to PvP, support more currencies and trade types, and reduce the large pool of transactions still settling gross bilaterally. At the same time, tokenised markets need legal finality, common standards and credible liquidity backstops.

For investors, the most important change may be invisible. Atomic settlement could shorten the period during which capital is exposed to a counterparty or trapped between cash and asset legs. That can improve capital efficiency and reduce some tail risks. But it does not make the asset safer, the counterparty solvent or the infrastructure infallible. Investors should distinguish settlement technology from the credit, market and liquidity characteristics of what is actually being settled.

The End of Settlement Risk? No—But Possibly the End of Avoidable Principal Risk

The promise of atomic payments is real because the problem is real. More than $1.4 trillion of FX obligations still settled each day in April 2025 without a mechanism that eliminated or mitigated settlement risk. At the same time, central-bank and industry experiments have moved beyond abstract demonstrations: Agorá has shown atomic multi-currency settlement, Meridian FX has linked different infrastructures, the Bank of England is testing synchronisation with industry, Australia has tested multiple forms of settlement money, and the ECB is preparing Pontes for launch.

What these projects collectively suggest is not the disappearance of settlement risk but its redesign. Principal risk can be engineered out of more transactions by ensuring that linked legs settle together. In exchange, institutions must manage more demanding liquidity, operational resilience, legal coordination and technology governance. The system becomes safer in one dimension and more dependent on infrastructure in another.

The most credible future is therefore neither a single blockchain nor a universal replacement for today’s market infrastructure. It is a financial system in which central-bank money, tokenised bank money and market assets can be synchronised across interoperable platforms with legally recognised finality. If that architecture scales, the phrase “settlement risk” will not disappear from risk committees. But one of its oldest and most avoidable forms—the possibility of paying away principal while the other side never arrives—could become much harder to justify.

References

1. BIS Quarterly Review, June 2026 — Uncovering FX settlement risk: new measures from the 2025 BIS Triennial Survey

2. BIS Innovation Hub — Project Agorá: a shared programmable platform for wholesale cross-border payments

3. BIS Annual Economic Report 2026, Chapter III — Anchoring trust in money: innovation beyond stablecoins

4. International Monetary Fund — Tokenized Finance, IMF Note 2026/001

5. BIS Innovation Hub — Project Meridian FX: exploring synchronised settlement in FX

6. European Central Bank — Pontes

7. European Central Bank — Focus Session: Preparing the go-live of Pontes, 22 July 2026

8. Bank of England — Extending RTGS and CHAPS settlement hours: next steps towards near 24x7 settlement

9. Reserve Bank of Australia — RBA and DFCRC Release Findings From Project Acacia

10. BIS Innovation Hub — Project Rialto: improving instant cross-border payments using central bank money settlement

11. Committee on Payments and Market Infrastructures — Facilitating increased adoption of payment versus payment (PvP)