Blockchain in banking and finance, from ledgers to programmable settlement

The financial system holds and creates bank money, channels savings into credit and investment, and maintains records that establish who owns an asset and who owes an obligation. Banks, central banks, markets, funds, intermediaries and depositories each perform different parts of that process.

Money itself operates in layers. The balance a person sees in an account is a debt their bank owes them. When they pay a customer of another institution, both banks update their ledgers and create an interbank obligation. Depending on the system, those obligations settle individually or after netting, normally in central bank money; cross-border payments may also involve correspondent accounts. A card payment or transfer starts an instruction and, behind it, several infrastructures check, clear and settle the movement.

Capital markets work in a similar way. A company or government issues a security; a market and its intermediaries make it tradable; a central counterparty may assume and clear the obligations; custodians and depositories maintain positions; and a payment system completes the cash delivery. Each layer exists to manage specific legal, credit and operational risks.

This architecture has allowed finance to operate at scale, but it does not behave as a single shared book. Every participant keeps its own version of the transaction and checks it against the others. Understanding where blockchain may add value begins with this combination of specialisation, messaging and reconciliation.

When an investor presses buy, an order may execute in milliseconds and appear immediately in an app. That agreement only fixes price, quantity and counterparties. The trade must still be confirmed and allocated, obligations calculated, securities made available, cash funded and delivery versus payment completed across the relevant infrastructures.

Trading happens on T. Clearing determines what each participant must deliver and may interpose a central counterparty that manages collateral and applies netting. Settlement comes afterwards. The depository moves the securities and the payment system moves the money until both transfers are final.

T+1 therefore does not mean an exchange takes a day to execute an order. For instruments covered by that standard cycle, final delivery must complete no later than the next business day. Government debt, funds, derivatives, foreign exchange, financing transactions and primary issuance may follow different timetables.

The United States shortened the standard cycle for most broker-dealer transactions to T+1 in May 2024. Mexico and Argentina adopted that timetable in specified cash-market segments. The European Union, including Spain, remains on T+2 during 2026 and is scheduled to move to T+1 on 11 October 2027.

Latin America does not have a single settlement cycle and cannot accurately be described as a T+3 region today. Mexico and Argentina already settle the specified segments on T+1, while Brazilian equities remain on D+2 and B3 plans a move to D+1 for February 2028. Other countries, instruments and trade types retain their own timetables.

A shorter cycle reduces the period in which a counterparty can fail and can release collateral earlier. In return, allocation, confirmation, securities lending, FX execution and funding are compressed into fewer hours. Elapsed exposure decreases while intraday funding and operational demands increase.

The interval is not simply a technical wait. During those hours, intermediaries, custodians, central counterparties, depositories and cash banks turn an execution into obligations that can be fulfilled. Each maintains its own record and receives instructions that must match on asset, quantity, account, date and currency.

A cross-border trade adds time zones and cycles that may not align. A European fund buying US securities on T+1 may need dollars before the sale of another T+2 asset settles. If securities are not available, it must borrow them; if an instruction does not match, there is less time to correct it.

The move to T+1 shows that conventional infrastructure can accelerate settlement. DLT addresses another friction by reducing how often the same state is replicated and reconciled, while allowing assets, money and operating conditions to be coordinated from a shared ledger.

A bank balance is a bank's debt to its customer. A reserve is a central bank's debt to an institution. A share or bond is a position recognised by an issuer and its market infrastructure. The financial system updates records of who holds a right and who carries an obligation.

Messaging, clearing, custody and settlement currently live in different systems and operating windows. A trade may be agreed and cleared through a CCP while the asset and cash still travel through separate chains. Cross-border payments add currencies, correspondent banks, controls and prefunded liquidity.

By design, DLT lets authorised institutions operate on shared state and execute rules over it. The gain does not come from calling an entry a token or replacing an intermediary with code. It comes from reducing duplication and coordinating the asset, money and conditions without losing governance, privacy or legal finality.

Technology can make different instruments look similar in a wallet, but the issuer determines the right and the risk. A tokenized deposit remains a bank liability. A stablecoin is a claim against its issuer and reserve structure. A CBDC is a central bank liability. A tokenized bond still needs a cash leg to settle.

This distinction determines who is liable, how redemption works, what happens in insolvency and which supervisor is involved. It is why an institution cannot choose the token before choosing the legal form of the money or asset.

Recording a transfer in seconds is not enough. Settlement requires the asset and an accepted form of money to be available, delivery to be linked to payment and the outcome to be legally final. If one of those pieces remains off-network, a bridge and reconciliation are still required.

Atomicity reduces the risk that one party delivers while the other does not, but it does not remove issuer credit, liquidity shortages, smart-contract failures or reliance on custodians, oracles and external networks. Gross transaction-by-transaction settlement may also consume more liquidity than a netting system.

A network may be available around the clock and confirm transfers in seconds, but a market does not have to settle every trade immediately on a gross basis. Waiting for a window allows offsetting buys and sells to be netted, reducing movements and the cash and securities required. The trade-off is exposure until settlement.

Atomic settlement links delivery and payment so both legs execute together or neither does. This reduces principal risk but requires the asset and an accepted form of money to be available on compatible infrastructure. If cash arrives on another network or only during central-bank operating hours, atomicity is constrained by that bridge.

DLT provides the capacity to settle earlier, conditionally or beyond traditional windows. The economic choice may still be to net and settle later. Serious infrastructure design must consciously balance counterparty risk, liquidity consumption, instruction volume and the ability to resolve exceptions.

Project Agorá brings together eight central banks and more than forty financial institutions. After publishing prototype results in May 2026, it completed real-value testing in July. The BIS describes multicurrency transfers with atomic settlement among participants, an experimental phase that does not yet constitute a generally available commercial service.

The Eurosystem settled more than 200 transactions worth €1.59 billion in its 2024 trials with 64 participants. Building on that work, it launched the first version of Pontes on 21 September 2026, a bridge between DLT platforms and TARGET for central-bank-money settlement, and is working on Appia, a longer-term track for a tokenized wholesale ecosystem.

In September 2026, J.P. Morgan reported approximately $7 billion in daily Kinexys volume and described foreign-exchange settlement across eight currencies. These are operator-reported figures for its institutional platform, rather than a measure restricted to international payments or evidence that banking networks are generally interoperable.

In capital markets, the EIB and other issuers have placed digital bonds inside the existing financial framework. Issuance may be native to DLT, while distribution, custody, cash, investor disclosure and trading still require complete market infrastructure.

Blockchain adds value where several institutions need to write to shared state, coordinate money with assets or preserve transferable rules across platforms. An application should reduce reconciliation, enable programmable settlement or improve instrument portability. Where one institution controls the whole process, a conventional database is usually sufficient.

Every form of tokenized money retains an issuer and a balance-sheet position. A tokenized deposit remains an obligation of the bank; a stablecoin depends on its issuer, reserves and redemption claim; wholesale central bank money is available only to eligible participants. A shared technical format does not make those risks equivalent.

Programmability changes when and under which conditions a position moves, but it does not create solvency or liquidity. If customers transfer tokenized deposits between banks, the issuing institution may need to fund the outflow and settle against another bank. If the market operates 24/7 while reserves or backing assets do not, an operating mismatch appears.

Design must anticipate concentration of balances, par convertibility, intraday limits, remuneration, accounting treatment and collateral availability. A solution can be technically atomic while creating a greater liquidity need than the netted process it replaces.

A trial may show that two tokens can be exchanged. Financial infrastructure must keep operating when a key is lost, an order is cancelled, an account is frozen, a coupon is miscalculated or a network stops producing blocks. The real scope includes the entire lifecycle and coexistence with legacy systems.

Production requires decisions on who performs each function, which record prevails, how changes are authorised and what happens in insolvency. It must also feed accounting, reporting, tax, statements and supervision without creating a second manual ledger around the network.

A settlement test should include lost communication after an instruction is submitted. Retrying without checking state may duplicate the operation; assuming failure may leave an unreconciled position. The business identifier should accompany the message, funds reservation and final confirmation. Operations staff need to inspect that journey without depending on the contract development team. Acceptance cases should cover delayed confirmation, counterparty rejection and recovery after interruption, demonstrating that the recorded outcome agrees with the actual obligation and cash movement.

Closing reconciliation compares obligations, available balances and accounting entries. If one network operates continuously while another has service windows, the design must explain when a position is funded and when it becomes usable. Value is measured through controlled completion, tied-up capital and outstanding exceptions. Faster consensus offers limited benefit when identity checks, limits or access to the payment system still block the complete journey. The operating model should make those boundaries visible to treasury and control teams.

A public network offers broadly accessible, composable state, while an institution still needs to control who may hold or transfer the instrument. A permissioned network simplifies privacy and governance but may recreate a silo if it cannot connect to other infrastructure.

Institutional designs are often hybrid. Identity and sensitive data remain off-chain, while rules and positions sit on a network with authorized participants and bridges to central bank money, deposits or other networks. The design depends on where each risk resides and who has authority to resolve an exception.

In the European Union, MiCA governs crypto-assets not already covered by other financial legislation, including e-money tokens and asset-referenced tokens. Tokenized bonds, shares and fund interests remain subject to financial-instrument rules; the DLT Pilot Regime permits certain trading and settlement infrastructures to operate under controlled exemptions.

MiCA does not turn every stablecoin into bank money and does not govern tokenized securities. Issuance, custody, order execution, e-money, payments, anti-money laundering, operational resilience and data protection may trigger separate obligations.

In the United States, the GENIUS Act became law in July 2025 and established a framework for permitted payment-stablecoin issuers, one-to-one backing, redemption and reserve disclosure. Tokenized securities remain subject to federal securities laws. Changing the record format does not by itself alter the instrument's legal nature.

In every jurisdiction, finality, insolvency, asset segregation and authority to correct the ledger must be settled before technical deployment. Programmable compliance executes controls; it does not decide what the law requires.

Blockchain can reduce reconciliation and support new execution patterns, but it does not create liquidity, solvency or legal finality by itself. Some risk moves into the design of money, code, key custody, external data and network connections. A production assessment must cover the entire operating lifecycle.

Tokenized financial infrastructure combines concepts from payments, securities markets and blockchain networks. Distinguishing messaging, clearing and settlement prevents a technical transfer from being mistaken for effects that depend on the money, legal record and finality rules. This vocabulary also separates deposits, stablecoins, CBDCs and tokenized securities.