The framing most people use here is wrong, and it creates unnecessary confusion. Banks aren't interested in blockchain because they've had some philosophical conversion about decentralization or peer-to-peer finance. If anything, decentralization is exactly what banks don't want. Their interest is narrower and more pragmatic: they have specific operational problems, and blockchain's properties — not blockchain's ideology — happen to solve some of them.
That's worth sitting with for a moment, because it changes how you interpret every headline about "Bank X explores blockchain." The question isn't whether banks have embraced crypto. It's which problems they're trying to fix, and whether blockchain actually fixes them.
The Settlement Problem Banks Actually Have
The most concrete driver is settlement lag. When you buy a stock on a public exchange, the trade executes in milliseconds. But ownership doesn't formally transfer until two business days later — what's known as T+2 settlement. The gap exists because the back-office infrastructure connecting brokers, custodians, clearinghouses, and registrars was built in layers over decades, and reconciling all of it takes time.
This lag isn't just inconvenient. It creates counterparty risk — the two-day window during which either party could, in theory, fail to deliver. Financial institutions have to set aside capital against this risk. The Depository Trust & Clearing Corporation (DTCC), which settles roughly $2.5 trillion in securities daily in the US, calculates and collects margin from member firms precisely because of this gap.
A shared ledger that updates simultaneously for all participants eliminates the reconciliation problem. If a buyer's custodian and a seller's custodian both write to the same ledger at the moment of trade, T+2 collapses toward T+0 or T+minutes. The DTCC ran a proof-of-concept on this architecture — Project Ion — before announcing a move toward accelerated settlement more broadly. The SEC finalized T+1 settlement rules in 2024 (effective May 2024), which is a step in this direction using conventional infrastructure. T+0 would require more fundamental plumbing changes.
Correspondent Banking Is Inefficient in Ways Blockchain Addresses Directly
Cross-border payments surface a related but distinct problem. Today, if a bank in Lagos needs to send dollars to a bank in Manila, it usually can't do so directly. It routes through one or more correspondent banks — intermediaries that hold accounts on behalf of smaller institutions globally. Each hop adds fees, delays, and opacity. A wire that should take minutes takes days, and neither the sender nor the recipient has visibility into where the funds are during transit.
SWIFT messaging coordinates the process, but SWIFT messages aren't money — they're instructions. The actual funds move through a patchwork of nostro/vostro accounts (essentially accounts banks maintain for each other to pre-fund cross-border transactions). Each bank needs to hold idle capital in foreign currencies to service outgoing transfers, which is expensive.
Blockchain-based settlement solves this differently. JPMorgan's JPM Coin — more precisely, a deposit token running on their private Onyx blockchain — allows JPMorgan's institutional clients to move dollar-denominated value across borders in near real-time, without the correspondent chain. The "coin" isn't a cryptocurrency in the retail sense; it's a tokenized bank deposit that represents a claim on JPMorgan and settles on their permissioned ledger. The mechanism is settlement finality on a shared ledger rather than message-passing through intermediaries.
Citibank, HSBC, and several others have similar projects at various stages. The Bank for International Settlements has run multiple multi-bank pilots — Project mBridge chief among them — exploring whether central bank digital currencies on shared infrastructure could replace correspondent chains entirely for wholesale transactions.
Smart Contracts as Programmable Compliance
There's a third driver that gets less attention but matters operationally: smart contracts can encode conditional logic that currently lives in legal agreements, compliance processes, and back-office workflows.
Trade finance is the clearest example. A letter of credit — the standard instrument for international goods transactions — involves a buyer's bank promising to pay a seller's bank once certain conditions are met (goods shipped, documents verified). The process is heavily manual, document-intensive, and prone to fraud through document falsification. A smart contract that triggers payment automatically upon verified delivery data from a shipping oracle doesn't just speed up the process — it reduces a class of fraud that's currently expensive to prevent.
Securities repos (repurchase agreements, where banks lend cash against collateral overnight) are another case. The terms are standardized, the collateral is trackable on a ledger, and the repayment is deterministic. Goldman Sachs' Digital Asset Platform has executed repo transactions on-chain with Goldman as one leg and counterparties on the other. The operational efficiency claim is that automated settlement reduces manual intervention, margin disputes, and reconciliation overhead.
Admittedly, this is still early. The smart contracts running on bank-operated chains are simpler than Ethereum DeFi contracts, and the legal enforceability of on-chain execution isn't fully settled across jurisdictions. But the direction is clear: programmable settlement beats manual settlement on operational cost.
Why It's Private Chains, Not Public Ones
Banks almost universally use permissioned blockchain infrastructure — private chains like Hyperledger Fabric, Corda, or their own proprietary variants — rather than public chains like Ethereum. This isn't incidental.
Banks need finality. A public chain transaction can theoretically be reorganized; a permissioned chain with known validators can provide settlement finality in seconds with no ambiguity about who's responsible. Banks also need compliance — they can't publish client transactions on a public ledger. They need control over who can participate. And frankly, they need the ability to reverse mistakes, which is impossible on an immutable public chain.
The irony critics point out is that if you remove decentralization, censorship-resistance, and permissionlessness from a blockchain, you're left with something close to a distributed database. That's a fair critique. The counterargument is that the specific properties banks want — shared append-only ledger, cryptographic audit trail, programmable logic, simultaneous update for multiple parties — are genuinely useful even in a permissioned context, and off-the-shelf database solutions weren't designed with multi-party simultaneous settlement in mind.
What Would Confirm This Direction
Confirmation signals worth watching: T+0 settlement becoming standard in major equity markets, requiring shared ledger infrastructure. Significant volumes moving through deposit token rails (JPM Coin and equivalents). Multi-CBDC interoperability networks (Project mBridge, or successors) reaching production. Reduction in nostro/vostro capital requirements at major correspondent banks.
What Would Break It
Invalidation signals: Settlement speed improvements achieved through upgraded conventional infrastructure (SWIFT Go, faster netting systems) without requiring blockchain. Major on-chain fraud event or smart contract failure causing reversal pressure. Regulatory prohibition on deposit tokens or bank-issued digital instruments. Failure of multi-bank blockchain consortia due to governance disputes — something we've already seen with R3's membership attrition.
Timing Perspective
Now: Wholesale deposit token systems are operational in limited production at several large banks (JPMorgan, Citi). T+1 settlement is live in US equities markets. Correspondent bank blockchain pilots are active.
Next: Whether T+0 settlement gains regulatory and infrastructure support over the next 2-3 years is the key question. Multi-CBDC systems reaching production would accelerate the correspondent banking displacement story.
Later: Whether public chains (via privacy layers like ZK-proofs) eventually serve institutional settlement needs, or whether the permissioned chain world remains separate.
What This Doesn't Mean
Bank interest in blockchain doesn't signal that banks are endorsing crypto markets, Bitcoin, or decentralized finance. They're solving specific back-office problems with specific technical properties, and the overlap with the retail crypto ecosystem is limited. A bank settling securities on a private Hyperledger network is operating in a different system than someone trading on Coinbase.
The questions worth tracking are operational: Are the efficiency claims holding up in production? Is the counterparty risk actually reduced? And does shared ledger infrastructure outperform the upgraded conventional systems that SWIFT and clearinghouses are also building? Those are empirical questions, and the answers are still coming in.
