tokenization

The Invisible Plumbing: How Tokenization Actually Works at the Settlement Layer

Block and Bond

— 8 min read
Share

Every bond trade has two lives. The first is the deal itself — the negotiation, the price, the handshake (or its electronic equivalent). The second is settlement: the quiet, largely invisible process by which that agreement becomes a legal reality. Securities move. Cash moves. Records change hands. The trade, as a matter of law, is complete.

Most of the innovation conversation in fixed income focuses on the first life — execution, pricing, distribution. The second life, settlement, is where the structural complexity actually lives. And it is in this second life that tokenization is beginning to matter in ways worth examining carefully.

What Settlement Actually Is

Settlement is the process of discharging the obligations created by a trade. In its simplest form, it involves four things: trade matching (confirming that both counterparties agree on what was transacted), confirmation (the official affirmation of trade details), delivery versus payment (the simultaneous exchange of securities and cash to eliminate principal risk), and the transfer of custody records to reflect new ownership.

None of this is glamorous. All of it is critical. A failure at any stage — a mismatch on settlement date, a shortfall in available securities, a delayed cash leg — creates a "fail," a trade that does not settle on time. Fails generate costs: regulatory penalties under frameworks like CSDR in Europe, financing charges, administrative overhead, and counterparty friction. In large portfolios, fail rates and reconciliation backlogs are a constant operational tax.

The delivery-versus-payment mechanism deserves particular attention. DvP is the principle that cash and securities should move simultaneously, eliminating the risk that one party delivers and the other does not. In practice, "simultaneous" in traditional markets means "within the same settlement cycle" — which is to say, not truly simultaneous at all, but rather close enough that risk is managed through central counterparties and collateral frameworks rather than genuine atomicity.

How Traditional Bond Settlement Works

In the United States, corporate and government bond settlement flows through the Depository Trust & Clearing Corporation's Fixed Income Clearing Corporation (FICC). In Europe, Euroclear and Clearstream serve as the primary international central securities depositories (ICSDs), with national CSDs handling domestic markets. These institutions hold the legal register of beneficial ownership, net multilateral obligations, and finalize settlement — typically on a T+2 basis, though some markets have moved to T+1.

The process between trade and settlement is more laborious than it appears from the outside. Counterparties must affirm trade details through separate affirmation workflows. Custodians must receive settlement instructions. Securities must be located, often requiring movement between sub-custodians or securities lending recalls. Cash must be positioned. Each of these steps involves at least one, often several, intermediaries maintaining their own records — records that must agree with each other and with the CSD by the time settlement is due.

This reconciliation burden is not trivial. It is a genuine cost center for every major custodian bank and asset manager. Systems that attempt to electronically match and affirm trades exist across the industry, but they operate as additions to the existing plumbing rather than replacements for it. The underlying settlement infrastructure — the CSD ledger, the RTGS cash system — remains centralized, fragmented across jurisdictions, and largely unchanged in its core architecture since the 1990s.

Fail rates in European bond markets have run consistently in the 5–10% range by value, generating penalties under the CSDR Settlement Discipline Regime. The causes are structural: settlement windows that do not accommodate late-day trades, insufficient securities inventory, and coordination failures between multiple custodians in cross-border transactions. T+2 is not a feature; it is a legacy constraint that the industry has built infrastructure around rather than resolved.

What Tokenized Settlement Changes

Tokenization of bonds means representing the security as a digital token on a distributed ledger — whether a public blockchain or a permissioned enterprise DLT. When both the security and the cash leg are represented on the same ledger, or when the ledger can coordinate atomically with a cash settlement system, a structurally different model becomes possible: genuine delivery versus payment, where both legs settle in the same transaction, at the same moment, with no interval of principal risk.

This is not an incremental improvement. Atomic DvP eliminates the exposure window that central counterparties exist to manage. It does not rely on netting, because there is no accumulation of bilateral risk to net. It does not require a separate cash leg affirmation, because the conditions for settlement are encoded in the transaction itself. Smart contracts can enforce settlement conditions — correct ISIN, correct quantity, correct counterparty, authorized by the relevant parties — without relying on bilateral legal agreements and separate operational workflows to ensure compliance.

Programmable securities take this further. A tokenized bond can carry its terms on-chain: coupon payment dates, redemption conditions, event of default triggers. Coupon payments can execute automatically against a register of token holders. Corporate actions — calls, exchanges, consent solicitations — can be coordinated through on-chain governance mechanisms rather than through the current process of instruction chains flowing through custodians and CSDs. The security and its administrative lifecycle exist in the same data layer.

On-chain custody records replace the reconciliation problem with a shared state. Rather than multiple institutions maintaining separate ledgers that must be periodically reconciled, there is one authoritative ledger to which all participants have read access. This does not eliminate the need for institutional participants, but it does change what they do: from maintaining proprietary records to operating nodes, managing keys, and providing legal and compliance wrappers around on-chain positions.

Where the Custodian Still Fits

Here is where enthusiasm about on-chain settlement often outpaces the structural reality. Cryptographic ownership and legal title are not the same thing. Holding a private key that controls a token representing a bond does not, on its own, confer legal title to that bond in any jurisdiction's securities law. Legal title depends on the applicable legal framework — whether that is the governing law of the instrument, the law of the place of incorporation of the issuer, or the law of the jurisdiction of the CSD.

Most tokenized bond frameworks to date have addressed this through a legal wrapper: the on-chain token is linked to a legal instrument governed by conventional securities law, and a licensed custodian or trustee holds legal title on behalf of token holders, or alternatively the relevant CSD acknowledges the token as the authoritative record of ownership. Neither approach makes the custodian unnecessary. It repositions the custodian as the bridge between cryptographic and legal ownership rather than the primary record-keeper.

Regulatory licensing requirements reinforce this. In virtually every major jurisdiction, holding client securities in custody requires authorization. In Europe, this means MiFID II, CSDR, and applicable national regimes authorization as a credit institution or investment firm. In the US, it means qualification as a broker-dealer or trust company. A smart contract cannot hold a custodial license. The institutional participants who hold those licenses are not going away; they are adapting their role within the new infrastructure.

This distinction matters. The argument for tokenized settlement is not that it removes institutional intermediaries. It is that it reduces the redundancy of institutional record-keeping, compresses settlement timelines, and makes the operational process less dependent on bilateral coordination. The intermediaries remain; the architecture around them changes.

Real Examples Worth Examining

The European Investment Bank has issued three digital bonds that illustrate the evolution of this space with useful specificity. In April 2021, the EIB issued a €100 million two-year digital bond (ISIN: XS2353983209) on the public Ethereum blockchain, co-managed by Goldman Sachs, Santander, Société Générale, and BNP Paribas — notably, settlement involved Banque de France CBDC experimentation rather than full production atomic DvP. Settlement was handled via a cash token issued by Banque de France, representing central bank money on-chain — a significant detail, because it meant the cash leg of the transaction was also on-chain, enabling genuine DvP. In November 2022, the EIB followed with a €100 million digital bond (ISIN: XS2577640649) on a private DLT platform, demonstrating that the same structural principles apply in a permissioned environment with different privacy and governance tradeoffs. In October 2023, a further EIB issuance (ISIN: XS2641028906) used Goldman Sachs's Digital Asset Platform and the Canton Network (a privacy-enabled interoperable DLT, not a single unified chain), a permissioned blockchain designed for institutional finance, with a particular focus on interoperability between institutional participants.

These three issuances are not marketing exercises. They are production transactions with real legal status, real settlement, and real ISINs registered with the relevant authorities. They demonstrate that the technical and legal framework for tokenized bond issuance and settlement is functional — not theoretical.

BlackRock's BUIDL fund — the BlackRock USD Institutional Digital Liquidity Fund, issued on Ethereum — has grown to approximately $2.54 billion in assets under management as of early 2026. BUIDL is a tokenized money market fund, not a bond, but its settlement mechanics are directly relevant: transfers between qualified investors settle on-chain, nearly instantaneously, with no T+1 or T+2 window. The fund demonstrates institutional-scale appetite for on-chain settlement in a fully licensed, regulated structure — BlackRock is the issuer, BNY Mellon is the custodian, and the on-chain token is the authoritative ownership record.

In March 2026, the Bank of Canada released findings from its Project Jasper, a DLT-based bond settlement experiment conducted with major Canadian financial institutions. The experiment demonstrated that atomic DvP settlement of government securities was achievable on a permissioned DLT, with settlement finality in seconds rather than days, and without the operational fails that characterize conventional settlement. The Bank of Canada was explicit that the experiment was exploratory — no production deployment has been announced — but the results confirmed that the settlement mechanics work at the level of sovereign debt instruments.

The Honest Assessment

Tokenized settlement changes specific, real things. Settlement finality compresses from days to seconds in the cases where both legs can be on-chain. Reconciliation overhead decreases when there is a shared ledger rather than multiple proprietary records. Programmable coupon and redemption mechanics reduce manual intervention in the bond lifecycle. Atomic DvP eliminates the principal risk interval that CCPs currently manage. These are structural improvements, not marketing claims.

What does not change, at least not yet, is the institutional landscape. The custodians, the CSDs, the transfer agents, and the trustees are all still present in tokenized bond structures. In some cases they are doing exactly what they did before, with a new technology layer underneath. In others, their role is genuinely different — operating nodes, managing keys, providing legal wrappers — but they are still there, still licensed, still charging fees.

The more honest framing is not "tokenization removes the intermediaries" but "tokenization changes what the intermediaries do and may reduce how many of them are needed." If the settlement function is automated on-chain, the value of maintaining proprietary settlement infrastructure diminishes. If the custody record is on a shared ledger, the value of maintaining proprietary record-keeping diminishes. Intermediaries whose value proposition depends primarily on information asymmetry — on being the authoritative record holder — face structural pressure. Intermediaries whose value proposition depends on legal authorization, client relationships, and regulatory compliance face much less.

The settlement layer is being rebuilt. Not quickly, not uniformly, and not without the existing participants firmly embedded in the new architecture. But the direction is clear, the production examples are real, and the structural logic is sound. For institutional fixed income, this is infrastructure evolution worth watching closely — not because it changes everything overnight, but because the market infrastructure it may ultimately replace are older than most of the people who maintain them.

Sources & References

Read more