Markets & Infrastructure · Tokenised Securities
Abstract
Tokenised securities are scaling while the necessity case for their underlying technology has failed. A companion analysis (Blockchain & Crypto Assets: A Critical Appraisal) concluded that almost every proclaimed advantage of the blockchain dissolves against the counterfactual of a conventional database, a trusted intermediary, or existing law. This essay treats the contradiction as a diagnosis rather than a paradox: adoption is being driven by institutional forces — collateral economics, balance-sheet metrics, licensing moats, and infrastructure politics — that never depended on technological necessity in the first place.
Five claims follow. The binding constraint on tokenised securities is economic, not technical: the wrapper has no natural payer, and it spreads only where the payer captures the gain. The cash leg has already been decided by the official sector in favour of the two-tier monetary system. The genuine efficiencies sit in collateral mobility and funding term granularity, not in retail access. The end-state is re-intermediation: central securities depositories and custodian banks will absorb the chain world, as Switzerland’s merger of its digital exchange into its legacy depository already demonstrates. And the new momentum behind the European single market is partly an infrastructure play — because a settlement stack cannot amortise across twenty-seven property-law regimes. Sections II, V, VI, VII and IX carry the five claims in turn; IV and VIII supply the operational evidence.
Quo vadis is a question you ask someone who is already walking. It is worth being precise about where the walking is actually happening, because the honest inventory of 2026 looks very different from the prospectus version.
In the European Union, the DLT Pilot Regime — the sandbox that was supposed to prove the model — produced a review before it produced a market. ESMA’s Article 14 report found modest uptake and identified the blockers with unusual candour: set-up costs, lack of interoperability with existing infrastructure, restricted access to central bank money for the cash leg, and uncertainty about whether the regime would survive at all (ESMA, 2025). The Commission’s response, tabled on 4 December 2025 within its markets-integration and supervision package, points towards permanence: time limits removed, the aggregate cap raised to €100 billion, a simplified authorisation tier for infrastructures below €10 billion in recorded instruments, and — more consequentially — amendments opening the CSD Regulation (CSDR) so that core depository services can be provided on distributed ledgers (European Commission, 2025). Trilogues are expected to run into 2027. The Eurosystem, meanwhile, is delivering the missing cash leg: Pontes, linking market DLT platforms to TARGET Services so that tokenised transactions settle in central bank money, had its initial launch on 21 September 2026 (ECB, 2025).
The other two reference jurisdictions moved in the same direction, which is to say: inward. In the United States, the SEC staff issued no-action relief in December 2025 for a tokenisation pilot at the Depository Trust Company — the incumbent depository, not a challenger (SEC, 2025) — while the GENIUS Act of 2025, the SEC’s September 2026 Innovation Exemption order opening conditional tokenised trading of listed equities (SEC, 2026), and the Commission’s declared tokenisation agenda point the same way: regularisation through the existing perimeter, not around it. And in Switzerland, the digital exchange was folded into the incumbent depository, creating the world’s first hybrid CSD — of which more in Section VII (SIX, 2026).
Against this institutional consolidation stands a market whose headline numbers deserve one deflationary footnote. Public dashboards of tokenised real-world assets (RWA) aggregate two categories that should never be added: assets that are merely represented on a ledger and cannot leave the issuing platform, and tokens that are genuinely distributed, transferable peer-to-peer. The leading dashboard now draws the distinction itself, toggling its aggregates between the two; as of late September 2026 the distributed aggregate stood at roughly $38.5 billion excluding stablecoins, against some $360 billion in the represented column — tokenised Treasury and money-market products the largest distributed category, and much of the retail-facing equity wave sitting on the represented side (rwa.xyz, 2026). A tokenised asset that cannot leave its platform is a register with marketing attached, not a market. Composition tells the same story: the largest live categories are tokenised money-market and Treasury funds — cash management wearing a token — and private credit, where tokenisation genuinely improves record-keeping and enables tranche-level transfer but manufactures no liquidity for the underlying loans (see The Loan Is Not the Pool).
One distinction keeps everything that follows clean, because the market is quietly separating three propositions that were sold as one. Tokenisation is the representation of a financial claim in programmable digital form. DLT is one way of maintaining and transferring that representation. Decentralisation is a governance choice. The first is proving economically useful; the second is operationally convenient; the third is where the original necessity claims lived — and where the counterfactual, argued in the companion appraisal, remains strongest. Nothing in this essay requires the third to succeed.
Read together, the pattern of 2026 is consistent everywhere you look: the institutional form of progress is the incumbent absorbing the technology. The rest of this essay is about why that was always the likely outcome, and what it implies for the questions practitioners actually face — corporate actions, native issuance, collateral, the cash leg, distribution, and the strange new energy behind the European single market.
Begin with the dominant structure in the market today: the wrapper. A security is issued and held conventionally — global certificate in a depository, or entries in a registrar’s book — and a token is created that mirrors it. Every wrapper carries dual running costs. The registrar does not go away; the ledger is added to it. Custody does not go away; wallet infrastructure is added to it. One reconciliation universe becomes two, and a set of legal opinions must be procured for every combination of chain, register and jurisdiction — plus the lines the brochures omit: key custody and HSM infrastructure, node operation, smart-contract audit and upgrade governance, and the operational-risk capital the new stack attracts. These costs are real, recurring, and land on the issuer, who passes them to the investor.
The efficiency gains, meanwhile, land somewhere else. Faster settlement, automated processing, reduced fails — these accrue largely to intermediaries and infrastructure operators rather than to the party paying for the wrapper. This incidence mismatch is the central economic fact of the market, and it explains the stall better than any technical limitation. It is the same structure as the T+1 migration: a systemic gain that no individual participant would fund voluntarily, which is why T+1 happened by regulatory mandate and socialised cost — a mandate that, for tokenisation, does not exist.
So who actually pays today? Three payers, each rational. Issuers and distributors pay where the wrapper is a client-acquisition cost: tokenised equity offerings aimed at retail are marketing budgets wearing market-structure clothing — and the quality of the claim inside the wrapper varies enormously (see The Inheritance Test). Banks pay where payer and beneficiary are the same balance sheet: internal collateral moves and intraday funding, of which more in Section VI. And the public sector pays as demonstration subsidy, the European Investment Bank’s digital bond programme being the canonical example.
The rule that falls out is simple and predictive: absent a mandate or a demonstration subsidy, the wrapper gets funded where the payer captures the gain. That predicts wholesale-internal use cases first, public-good demonstrations second, retail wrappers as loss-leaders, and — until mandate or bundling changes the incidence — nothing else at scale. The named failure mode of the entire sector is that as long as gains are intermediary-side and costs are issuer-side, tokenisation spreads only where it is bundled with distribution or ordered by regulation.
The obvious escape from dual costs is to stop wrapping: issue the security natively on the ledger, so that the token is the security. The statutes exist and have existed for years:
Native issuance solves the duplication problem and immediately creates the orphan-security problem. A natively issued instrument is born outside the ecosystem that gives a security its value in use. Until Pontes and its equivalents scale, there is no central-bank cash leg. Central-bank collateral frameworks tie eligibility to settlement through approved systems, so a native instrument outside them cannot — without first clearing a heavy burden of legal certainty, finality and custody compatibility — be pledged at the ECB or SNB. Index providers do not include it, so passive money cannot hold it. Standard repo documentation assumes conventional delivery mechanics, so it cannot easily be financed. Institutional mandates frequently require CSD-held assets, so the natural buyers are constrained out.
Beneath all of these sits the question the ecosystem discussions politely skip: legal finality. Under the Settlement Finality Directive, transfer orders in designated systems are protected against insolvency unwind — that protection is what makes a settlement final in law rather than merely irreversible in software (Directive 98/26/EC). An on-chain confirmation is operationally hard to reverse; it is not, by itself, legally final. A native instrument transferring on a ledger outside any designated system carries transfers that an insolvency administrator can, in principle, still attack — which is precisely the risk wholesale counterparties are not paid to take. The DLT regimes let pilot infrastructures obtain designation; everything outside them settles on hope and contract law. The December 2025 package even carries a dedicated Settlement Finality Regulation to extend that protection onto DLT — an admission, in legislative form, that the gap is real (European Commission, 2025).
A security that cannot be repo’d, pledged or indexed is a worse security, whatever its settlement speed. This — not legal uncertainty, which the statutes largely resolved — is why native issuance volumes remain demonstration-grade. The EIB’s digital bonds proved feasibility repeatedly; they could not prove ecosystem completeness, because no single issuer can.
The direction of resolution is visible, and it is not the one the early literature imagined. The Swiss hybrid depository and the EU proposal to let CSD services run on DLT both point the same way: native issuance will happen inside the incumbent perimeter, where the cash leg, collateral eligibility and index plumbing already live. The orphan gets adopted — by the family it was supposed to replace.
Ask practitioners what actually breaks in tokenised securities operations and the answer is rarely settlement. It is corporate actions — and the difficulty is structural, not transitional.
Draw the distinction properly. Mandatory, deterministic events — coupon payments, redemptions, splits — are arithmetic performed on a register. They are trivially automatable on a smart contract, and, as the companion appraisal noted, equally automatable on a database; automation here is real but not distinctive. Elective and discretionary events are a different species. Rights issues, tender and exchange offers, consent solicitations, proxy voting, dividend elections: these are options with deadlines, exercised by identified holders, communicated through defined channels, snapshotted for entitlement, and — crucially — capable of being contested. A dividend is arithmetic; a rights issue is a negotiation. Negotiations require identity, discretion and a finality that survives litigation, none of which a ledger natively carries.
Withholding tax makes the point empirically. Relief at source depends on beneficial-owner identity and treaty status, verified along a chain of intermediaries. The flagship corporate-actions reform the EU legislated this decade — the FASTER directive on withholding-tax relief, applying from 2030 — is built entirely on national registers, digital residence certificates and certified financial intermediaries (Council of the EU, 2025). The ledger plays no role in it. The reform of the very process tokenisation was supposed to streamline was designed, in the mid-2020s, without reference to tokenisation at all.
Beneath everything sits the golden-source question. If the chain is the legally definitive register, then an erroneous on-chain corporate action is legally definitive nonsense, and someone must own the machinery for reversing the irreversible. If the chain merely mirrors an off-chain register, divergence is possible, divergence creates liability, and reconciliation — the cost tokenisation promised to remove — returns through the side door. The current practical answer across the market is the second one: elective actions stay with the agent, off-chain, and results are written to the ledger afterwards. The token holds the claim; the lawyers hold the event.
What would genuinely solve this is a standardised on-chain entitlement and election messaging layer — an ISO 20022 equivalent for smart contracts, adopted across issuers and agents. Note who would have to build it: the corporate-actions agents whose fee base it would commoditise. Section II’s incidence problem, again, wearing different clothes.
A security settles against money, and delivery-versus-payment is only as good as its weaker leg. Three candidates compete for the money side of tokenised markets: stablecoins, tokenised commercial bank deposits, and central bank money made reachable on programmable rails. For regulated wholesale settlement, the competition is effectively over — not because a market chose, but because the official sector did; outside that core, stablecoins keep their specialised roles.
The BIS General Manager’s Jackson Hole speech in August laid out the selection criteria and the verdict with unusual directness (Hernández de Cos, 2026). The decisive argument concerns singleness. A holder of one stablecoin paying a counterparty who accepts another must route through a secondary market where deviations from par are the norm and sizeable under stress; nothing in the architecture enforces one-for-one exchange. For securities settlement the implication is mechanical: a settlement asset that can trade away from par is not a cash leg — it is a second risk position inside the trade. Add the integrity problem — recent evidence suggests the majority of stablecoin balances sit in self-custodied wallets, outside any KYC perimeter (Gopinath, 2026) — and stablecoins fail the wholesale test on microstructure and compliance grounds simultaneously, before regulation is even invoked.
The framework the official sector has converged on is therefore two-tier by construction: central bank money anchoring par settlement on tokenised platforms, tokenised deposits carrying the bulk of wholesale payments within prudential perimeters, and stablecoins relegated to specialised roles or reclassified as investment products (BIS, 2026a). Project Agorá has demonstrated cross-border feasibility for the tokenised-deposit model (BIS, 2026b); Pontes supplies the central-bank anchor in the euro area. The candour in the speech deserves note: no multi-bank, inter-jurisdictional tokenised-deposit ecosystem yet exists, and several current implementations are better described as bank-issued stablecoins. The architecture has been selected; it has not been built.
One regional correction to that framing, because an EU-facing essay cannot discuss the cash leg as if classification were still pending: MiCA has already answered the taxonomy question. E-money tokens and asset-referenced tokens are defined categories with par-redemption rights, reserve requirements and an interest prohibition, with significant tokens supervised at EBA level (Regulation (EU) 2023/1114). The “reclassified as investment products” scenario is the American debate; the European one was settled by legislation before the BIS speech was written — and settled, note, in exactly the direction the framework prescribes: stablecoins as regulated payment instruments inside a perimeter, not as free-floating settlement assets. The EU pre-committed to the two-tier answer.
Two caveats keep this honest. First, the BIS is talking its book — tokenised deposits preserve precisely the two-tier system whose stewards write BIS speeches. The correct reading is not that neutral analysis favoured incumbency, but that the institutions with authority over the cash leg have pre-selected the incumbent-compatible answer; for forecasting purposes, that is the more useful fact. Second, the macro-financial stakes are real rather than rhetorical: stablecoin reserve composition feeds directly into bank funding costs and credit conditions, and a fully-reserved stablecoin can become a safe haven that drains bank deposits precisely in stress (Hofmann, Kaldorf and Rottner, 2026).
The consequence for market structure follows immediately and matters for Section VIII: wholesale DvP in tokenised securities will settle in central-bank-anchored, permissioned money. Venues that cannot reach those rails are excluded from wholesale settlement by construction — no prohibition required.
If the retail story is marketing and the wrapper has no natural payer, where do the genuine efficiencies live? In the least photogenic corner of finance: secured funding and collateral management. This is the one segment where Section II’s rule is satisfied — the bank paying for the infrastructure is the bank capturing the gain — and, not coincidentally, the one segment that has actually grown.
The innovation is not speed. It is term granularity. Conventional repo has a minimum tenor of overnight: a bank facing a funding gap of a few hours borrows for twenty-four, over-mobilising collateral and carrying balance-sheet exposure for the nineteen hours it never needed. Intraday liquidity, meanwhile, is managed not with priced funding but with buffers, unremunerated credit lines and pre-positioned collateral, monitored under a supervisory framework that exists precisely because the exposure cannot be cleanly funded (BCBS, 2013). Tokenised repo, settling delivery-versus-payment at precise times of day, makes tenor a continuous variable. The gap itself becomes the instrument. (Where that leads — from tenor through balance-sheet capacity to the price of credit — is traced in The Shortening of Money.)
The killer app of tokenisation is not democratised ownership. It is a bank borrowing for four hours, twenty-one minutes and thirteen seconds — because that was the exact length of the gap.
The evidence base here is production-grade rather than pilot-grade: intraday repo has run on bank-operated ledgers since 2020 (J.P. Morgan, 2020), Broadridge’s Distributed Ledger Repo platform by itself processed $7.5 trillion in June 2026 (Broadridge, 2026), and collateral-swap infrastructure transfers ownership at precise moments without moving the underlying securities between custodians (HQLAx, 2019–2026). The balance-sheet economics are quantifiable: funding that exists only inside the gap and is extinguished before the reporting snapshot carries a different metric cost than overnight money; collateral velocity rises; buffers shrink. These are gains a treasurer can put a number on, which is why this segment did not need a marketing narrative.
The purest expression of the collateral case is not the repo trade at all but the margin transfer: posting a tokenised money-market fund share instead of cash. Margin held as cash earns little and must be raised by selling something; a tokenised MMF share moves as a registered claim — the fund’s assets never move, the yield never stops, and the transfer settles in minutes at any hour. This, more than any distribution story, is why tokenised money-market funds became the largest live RWA category (Section I) and why their registers sit at transfer agents rather than depositories (Section VII): the product is not a fund with a token attached but collateral engineered for mobility — and it thrives under Section V’s regime precisely because collateral does not need to be money, so a claim that may trade at a haircut passes a test the cash leg cannot. The disadvantages have not disappeared; they have changed address. Operationally there are none worth naming. Prudentially, three: acceptance — clearing-house and margin rulebooks still list cash and government bonds, so eligibility is a rulebook decision, not a technical one; stress behaviour — a fund share carries the fund’s redemption machinery, gates and liquidity fees included, which means the collateral is least reliable precisely when it is being called, the wrapper inheriting the fund’s stress liquidity as it inherits everything else; and valuation — a daily NAV posted against intraday calls prices as a conservative haircut. The gain is genuine; the residual risk has merely moved from the operations department to the rulebook and the stress scenario.
Two named failure modes bound the use case, and both deserve more attention than they get. First, there is no intraday yield curve. Every benchmark rate — €STR, SOFR — is a daily fixing; a four-hour repo has no reference price. Continuous tenor without a continuous benchmark means bilateral pricing, and bilateral pricing means dealer spread — so a portion of the efficiency gain leaks straight back to the intermediary. Section II’s incidence problem reappears inside the flagship use case. Second — and this is the strongest technical objection to the entire thesis, so it deserves its full weight — the real issue is settlement granularity: settle every trade atomically and individually, and the multilateral netting that does the system’s quiet work is forgone. Multilateral netting at central counterparties compresses gross obligations by orders of magnitude before anything settles at all; settle everything atomically, trade by trade, and every leg needs funding at full size at its own moment, so system-wide liquidity needs rise rather than fall. Payments learned this lesson decades ago, which is why pure real-time gross settlement gave way to hybrid systems with netting windows — and tokenised securities settlement will rediscover the same compromise, with batched or netted settlement cycles running on the ledger. Tokenised settlement therefore wins where trades are bespoke, collateralised and time-critical, and loses to netted clearing for high-volume flow. That is the honest boundary of the killer app: it is a precision instrument, not a replacement market.
History first. Dematerialisation — the last great re-platforming of securities — did not disintermediate anything. Paper certificates were immobilised, immobilisation required a vault, the vault became a book-entry system, and the book-entry system became the central securities depository: an institution that barely existed before the technology change and emerged from it holding legal privileges, insolvency protections and a licence moat. Dematerialisation did not abolish the depository; it invented it.
The question of who absorbs is usually asked as if depositories and custodians were one category. They are not, and the answer splits along function. The register function pools at the CSD by legal mandate: the CSD Regulation requires venue-traded transferable securities to be recorded in book-entry form at a depository, and the June 2026 reform answers the question of who may operate the ledger with “the depository, now permitted to run it on DLT” (European Commission, 2025). Switzerland reached the same answer by merger — the world’s first regulated digital exchange ended its institutional life in May 2026, absorbed with FINMA’s approval into SIX SIS, the legacy depository, as a single licensed entity spanning conventional securities, ledger-based securities and crypto custody — “one plug to two worlds,” in the group’s own phrase (SIX, 2026) — and the United States reached it by relief, tokenising at DTC rather than around it (SEC, 2025). Law has already assigned the notary layer to the incumbent registers.
The economics pool somewhere else. The value in tokenisation, as Section VI argued, sits in asset servicing and collateral mobility — and those are custodian businesses. The collateral networks are being built by the global custodian banks and dealer banks as product lines, not by depositories and not as new firms. More telling still is the largest live category of tokenised real-world assets, money-market and short-term Treasury funds: fund shares were never subject to the CSD book-entry mandate, and their register lives at the transfer agent. The flagship growth segment of the entire RWA market is therefore bypassing the depository layer altogether, absorbed instead by fund administrators and custodians running the register as a service. The world’s largest custodian made this concrete in July 2026, launching global digital transfer-agency capabilities with legal representation of the fund’s books and records held on a public blockchain (BNY, 2026b) — the register itself, absorbed as a product line, by the incumbent it was supposed to threaten. Where the ledger replaces the transfer agent’s database rather than the depository’s, the depository never enters the trade.
The notary layer goes to the depositories by mandate. The margin goes to the custodians by economics.
Two qualifications complete the picture. First, the categories blur exactly where absorption is happening: the international CSDs are banks holding depository licences, and the Swiss hybrid is custodian and depository in one entity — groups that are both absorb both, which makes the ICSDs the structurally best-positioned absorbers in Europe. Second, there is an escape hatch, and it matters for issuers: the book-entry mandate binds venue-traded securities, but private placements can be issued natively on registers operated by licensed non-CSD registrars — the German crypto-securities registrar is the cleanest example. The depository’s moat covers the public market; the private market can route around it, and that is where genuinely chain-native structures will accumulate.
The split is not entirely stable, and the instability deserves naming because it defines the infrastructure conflict of the 2030s. A custodian that issues tokenised deposits — as the largest began doing in January 2026, explicitly for collateral, margin and payments (BNY, 2026a) — and that custodies both sides of a trade can settle tokenised assets against tokenised cash atomically inside its own books. That is internalised settlement: the custodian operating as a de facto private depository for the perimeter it hosts, consuming no market infrastructure because it has become one. The attractions are real (integration, speed, a single legal environment) and so are the objections — these are precisely the proprietary silos the official sector has warned against (Hernández de Cos, 2026), with interoperability surrendered and settlement concentrated on a handful of commercial balance sheets. Nor is the pressure point hypothetical: settlement internalisation is already a reported, supervised category under the CSD Regulation, and at sufficient scale supervisors have historically forced internalisers back toward designated-system status. Whether a commercial bank’s ledger may be the depository of the tokenised world, or will be dragged inside the utility perimeter, is the genuinely open question this section’s tidy division leaves behind.
The mechanism throughout is not nostalgia; it is that absorption follows the licence and the liability ledger. Whoever operates the legally definitive register and holds the insolvency-remote structure absorbs any new recording technology, and the technology provider becomes a supplier. So the verdict is split, and precisely: the notary layer goes to the depositories by mandate, the margin goes to the custodians by economics, the one contested boundary is where internalised settlement ends and the depository’s mandate begins — and by 2030 “blockchain” appears in both product sheets roughly the way “SQL” does today: as infrastructure vocabulary, not as strategy.
If depositories and custodians absorb the plumbing, what remains contestable is distribution — and the two challenger classes face very different ceilings.
Centralised crypto exchanges are converging on the broker-bank model, a convergence visible in the wave of MiFID-licence acquisitions across 2025 and 2026 — Kraken and Coinbase each bought their way to an EU investment-firm licence through Cypriot acquisitions in 2025. What they win is real but bounded: the crypto-native client base, superior onboarding, and a round-the-clock user experience that traditional brokers will spend years matching. What they do not win is market structure. A licensed exchange distributing tokenised securities plugs into the settlement and cash-leg rails described above; it does not set them. The likely end-state is that the largest CEXs become what they already resemble — retail brokerages with custody arms — competing with incumbents on interface rather than infrastructure.
Decentralised exchanges face two walls, and it matters that neither is a prohibition. The identity wall: securities law requires knowing your holders — for prospectus liability, corporate actions, sanctions, and withholding relief — and a genuinely permissionless venue has no native way to reconcile pseudonymous participation with those obligations. The singleness wall, from Section V: the settlement asset available on public chains is the stablecoin, which fails the cash-leg test on microstructure grounds. Wholesale tokenised markets will settle in permissioned, central-bank-anchored money that public-chain venues cannot reach. The residual niche is genuine but narrow: permissioned pools operating inside licensed perimeters, where automated market making survives as an auction mechanism adopted by regulated venues — a technology transfer, not a market transfer. The distribution layer, in short, is where tokenisation is most visible and least consequential: the retail token is a client-acquisition instrument, and the register behind it is invisible to the client who holds it.
Beneath the venue question sits the one this essay has so far dodged: do tokenised securities actually trade? Mostly, no. Much of the tokenised bond market still trades by appointment, and the round-the-clock transferability of the wrapper is not liquidity — liquidity is made by market makers, market-making requires inventory, and inventory requires financing, which returns the problem to Section III: an instrument that cannot be repo’d cannot be warehoused, and an instrument that cannot be warehoused cannot be quoted. Twenty-four-hour markets, in current practice, mean twenty-four hours of exposure to thin books. The container does not manufacture the liquidity (see The Wrapper Fallacy), and neither does the venue. Liquidity will arrive where it always arrives — where instruments are standardised, financeable and index-included — which is to say, inside the incumbent perimeter of Section VII.
Why is Brussels suddenly serious about capital markets integration after three decades of stalling? The standard answers — capital flight to the US, defence financing needs — are true and incomplete. There is an infrastructure answer, and it inverts the usual causality.
A settlement stack is a fixed cost of enormous size. Fixed costs demand amortisation; amortisation demands scale; and scale, in securities, demands legal uniformity — which is precisely what the Union does not have. Article 345 TFEU shields national systems of property ownership from Union displacement, which is why every attempt to harmonise the substance of securities holding has died at the same altitude: the proposed Securities Law Directive was consulted on in 2010 and never became a proposal. The result is that the EU passports the document but not the security. A prospectus approved in one member state travels freely; the holding model, registrar and notary requirements, withholding procedures and national investor-protection add-ons do not. Issuers respond by running parallel structures per jurisdiction — and for tokenised issuance the multiplication is worse, because every combination of ledger, register statute and national property law needs its own legal opinion. The per-jurisdiction offering cost is not a friction in the system; under Article 345, it is the system.
Three escape paths exist. A twenty-eighth regime — an EU-level digital-securities statute, a Union Wertrecht sitting alongside national law — is the elegant answer, with a named failure mode: optional regimes historically underperform, because national gatekeepers price the default and the option withers (the European Company and ELTIF 1.0 are the cautionary precedents). Mutual recognition of DLT registers under a reformed CSDR is the unglamorous answer and the actual direction of the June 2026 proposal. And hub concentration is the market’s revealed preference: issuance migrates to one or two friendly jurisdictions and passports outward — the Luxembourg precedent, in which UCITS solved a harmonisation problem by turning one member state into everyone’s back office.
The technology is not waiting for the single market. The single-market case is being rebuilt to justify the infrastructure.
Hence the inversion worth stating plainly: the technology is not waiting for the single market — the single-market case is being partly rebuilt to justify the infrastructure. Tokenisation never earned its legislative attention on the necessity case, which failed. It earned it as industrial policy: the argument that if Europe does not build permissioned, euro-denominated rails, wholesale digital finance will settle on dollarised public ones. Monetary sovereignty is doing the argumentative work that efficiency could not. That is not a criticism; it is an identification of the actual driver, stated openly in the Eurosystem’s own speeches — and drivers, unlike narratives, predict budgets. It also closes the loop with Section II: what is arriving from Brussels is not the missing settlement mandate but an infrastructure subsidy at continental scale — the second of Section II’s named exceptions, industrialised.
No big bang, then. The sequence is already legible and runs in strict order of Section II’s rule. Internal bank plumbing first, because payer and beneficiary coincide — this is happening. Wholesale collateral and repo second, because the gains are quantifiable on the treasurer’s own book — this is growing. Primary issuance inside hybrid depositories third, as the CSDR amendments and the Swiss model make the incumbent perimeter ledger-capable — this is starting. Retail distribution wrappers fourth, funded as marketing for as long as marketing budgets last. Retail native issuance last, if ever: retail may be the largest distribution opportunity, but the retail investor captures too little of the infrastructure gain to finance the market-structure cost — and has no ability to price the claim she holds.
The apparent paradox of the field — a technology scaling after its necessity case failed — resolves once the drivers are named. Institutions are adopting tokenisation for institutional reasons: balance-sheet metrics, licence moats, sovereignty politics, distribution economics. None of these ever depended on the blockchain being necessary. They depend on it being there, and fundable, and absorbable.
Since the essay has been confident throughout, it should say what would prove it wrong. Three observables would falsify the absorption thesis: a large issuer running native issuance at benchmark scale outside any depository perimeter and finding repo, index and collateral acceptance anyway; a major jurisdiction permitting wholesale DvP settlement in a private stablecoin; a licensed venue running meaningful automated-market-maker volume in regulated securities; or — the direct test of the central claim — a token-native market infrastructure growing large enough that the incumbents connect to it rather than absorb it. None is visible in 2026. Watch for all four.
And absorbed it will be. The chain is on the same trajectory as every successful piece of financial plumbing before it: into the back office, owned by the incumbents, invisible to the end investor, wholly successful and wholly unremarked — the SWIFT trajectory, not the internet trajectory. Quo vadis? Into the depository. The revolution ends, as revolutions in finance usually do, with new plumbing under old law — and the invoice quietly forwarded to whoever captures the gain.
Written in a personal capacity. Analytical views only — not legal or investment advice.
Julian Gretzinger — independent adviser and regulatory strategist working across capital markets, tokenised securities and AI infrastructure policy in the Liechtenstein–Swiss corridor. He publishes at juliangretzinger.com and on Substack.