Tokenisation changes the medium — whose record makes you the owner is decided register by register, under national law
Tokenised securities are discussed as a matter of technology and venue: which chain, which market, what hours. Those questions decide little. The structural question is whose record legally constitutes ownership, and on that question much of what is marketed as tokenised changes nothing: it is an existing account position, or an existing register, written in a new medium.
This essay sorts securities by constitutive ledger rather than by technology, applies the sorting twice to every wrapped product, and runs fund units, bonds and listed shares through it. Tokenisation can end the intermediated title chain for newly issued instruments. It does not end intermediation, it does not escape national property law, and the layers likely to globalise are the custodial token and the bridge, not the register.
Take a tokenised security and remove everything that is said about it: the chain it runs on, the venue it trades on, the hours, the speed, the word "ownership" in the announcement. What remains is a record, and three questions. An entry in what? Kept by whom? Under whose law?
Every security is, in the end, an entry: a line in some record that the law treats as making someone the holder. For centuries the record was a certificate. Since the paperwork crises of the 1960s and 70s it has been, for most securities, a chain of accounts with a central depository at the top. Tokenisation is a proposal to change the record again. Its public discussion is about almost everything else: venues, trading hours, liquidity pools, settlement speed.
Put the three questions to the products themselves. Three exhibits from the past year, one for each kind of thing now called tokenised.
These are real engineering projects. They change how a position is written and moved. None of them changes who the registered holder is or what, in law, the position amounts to.
That is broadly how earlier changes of medium went. Locking certificates in a vault, and later abolishing them, generally changed the form of ownership where the law moved the constitutive record from the paper to the account, and left it alone where the law did not. Technology seldom did it alone, and there is little reason to expect a distributed ledger to be different. A new medium tends to become a new form of ownership when the law treats its entries as the record.
So the venue question, which the announcements answer, is the minor one. The major one is which record constitutes the right, who keeps it, and under whose law. Framed that way, the contest is narrower than the tokenisation narrative claims and more durable than its critics allow.
A word on scope, because it is part of the argument. Almost anything can be tokenised in the loose sense: securities, money claims, gold and commodities, buildings, receivables, carbon credits, art and collectibles. This essay confines itself to securities because they are the class whose ownership is constituted by a record that market participants keep — a registrar, a depository — so the token competes with an existing register, and the contest can be described. Land, where its register is constitutive, has a state-kept one that shares the seat with no one; there is nothing to contest. For everything else the prior questions come first: is the asset there, who holds it, whose is it. A token over a bar, a painting or a flat is a claim on whoever answers those questions, and its quality is the custodian's. My expectation, offered as a forecast rather than a finding, is that such tokens will develop and will trade — but as claims on a named custodian, and, where the asset is unique rather than fungible, bilaterally or inside the issuer's own platform rather than on public venues. A venue needs a fungible claim it can settle against a record it trusts; outside the register there is no such record, only the custodian's word. The venue presupposes the register.
First, how the present system works, since everything below is measured against it. A company or fund keeps a register of who holds its securities; the firm that keeps it for the issuer is the registrar, called a transfer agent in the United States. In the dominant model for listed securities that register carries little more than one name. The issue is lodged with a central securities depository, or CSD (below simply "the depository"), such as the Depository Trust Company (DTC) in the United States or Euroclear and Clearstream in Europe. The depository, or a holding entity it uses for the purpose, called its nominee, is the name on the issuer's register.
The depository does three things. It reconciles the securities credited through the system against the securities issued. It keeps accounts for its members, the large banks and brokers, showing how much of each issue each one holds. And it settles trades by moving securities between those accounts. The members in turn keep accounts for smaller banks and brokers, who keep accounts for investors. An investor's holding is therefore a credit in a broker's books, backed by a credit in a bank's books, backed by a credit at the depository. That chain exists because it is efficient: thousands of trades cancel out into a few net movements at the top, a saving known as netting.
The chain has a failure state, and Delaware has seen it. When Dole Food was taken private, 36,793,758 shares were entitled to a later settlement; claims from beneficial owners arrived for 49,164,415. Nobody had cheated — more than 32 million shares had been traded, lent and sold short in the final three days, so that the claims running through the tiers outran the shares on the depository's ledger. The court found the question of who owned the shares at closing unanswerable from the records, allocated the money through that ledger — the single name on the register — and left the tiers below to pass it on (In re Dole Food Co. Stockholder Litigation, Del. Ch., 2017). That is the system whose record tokenisation proposes to rewrite.
Sort securities by the record whose entries the law treats as constitutive: the record that, when it changes, changes the holder. Where a record is only evidence of a right constituted elsewhere, it is not the one to sort by. For the purposes of this essay three ideal types are enough. They are an analytical frame, not a legal taxonomy: each national regime fills them with its own doctrine.
Three observations follow. First, the direct register is not a blockchain invention. Registered shares and the US Direct Registration System are direct registers kept in ordinary databases. What statutes such as the Swiss DLT Act, the eWpG and the TVTG added is the possibility that the register is a distributed ledger and that a transfer on it is, without more, the transfer of the right.
Second, a direct register is direct only for whoever is entered on it. A nominee can stand on a perfectly constitutive register, and the investor behind it is back in an account chain. The question is always whose name the entry carries.
Third, native tokenisation exists only on a direct register. By native I mean, in this essay, that the legally operative register is itself kept on a distributed ledger and the governing law gives its entries that effect. A token can be the constitutive entry only where the law says the ledger it lives on is the register. Everywhere else the token is a pointer to an entry made somewhere else, and the pointer inherits every weakness of the path to that entry.
The frame looks too tidy the moment it meets a wrapper, and wrapped and custodial products are most of the market. The repair is simple and, as far as I can find, rarely done: a wrapped instrument is two instruments under one ticker, and each needs its own answer to the same two questions. Which ledger is constitutive? Which law governs it?
The word needs fixing first, because the market uses it for two structures that fail differently: a wrapper creates a second instrument; a custodial token lengthens the chain of the same one. A wrapper, in the narrow sense used here, is a second instrument from a second issuer: the holder has a claim against that issuer, and the issuer, not the holder, owns the underlying. A custodial token is something else: the same instrument, held for the investor by one more intermediary who writes its account entries on a chain. The first adds an instrument. The second adds a tier. Only the first needs the frame applied twice; the second needs it applied once, to a longer chain.
Take the wrapper proper: a tokenised share issued by a special-purpose vehicle. The token is a debt or tracker instrument of the vehicle. Its constitutive ledger may well be the chain, under the vehicle's home statute: a genuine direct register. The underlying share is held by the vehicle, or its custodian, as an account-chain position under New York law. The holder has direct, register-level title to a claim against an issuer whose own asset is an intermediated entitlement.
Two instruments, one ticker. The marketing describes the first. The risk lives in the second.
Applied twice, the frame gives each product a pair rather than a label. The pairs are ideal types, and real products combine them; the offering documents and the operative records decide which is which, never the ticker or the interface.
The corollary is uncomfortable for both camps. The tokenised equity models trading today are, with few exceptions, one of the first three: account-chain substance in direct-register clothing. The exceptions are issuer-sponsored arrangements in which the transfer agent's on-chain record is the shareholder register. Galaxy Digital's 2025 structure with Superstate is the usual exhibit: on the issuer's own description, the tokens are its common stock and the registered shareholder list updates as they move. Such cases are small, and they show that listed shares can sit on a direct register where the issuer wants it. The SEC's exemption requires that a tokenised stock carry the same rights as the share; it regulates the equivalence of rights and is silent on the identity of the ledger. Equal rights reached through a longer chain are not the same position in an insolvency — and the rights themselves are exercised through the chain: in a custodial-token model, voting and corporate actions still run through whoever keeps the entry. That is the ground covered by the Inheritance Test and by The Wallet Does Not Tell You; here it is only the premise.
If the register is the prize, the contest needs restating. It is usually told as incumbents against crypto. Properly stated it is a contest between two incumbent roles: the central securities depository, which sits at the top of the account chain, and the registrar or transfer agent, which keeps the issuer's own record. In an account-chain world the registrar's book has one line on it, the depository's nominee, and the registrar is a utility. In a direct-register world the registrar's book is the market, and the depository has to justify itself by what it does rather than by where it sits.
The custodian banks are the third party, and their interest is not what it appears. They do not need to win the seat. A wrapper built on their custody keeps the underlying inside the account chain, keeps asset servicing and collateral with them, and lets someone else carry the regulatory cost of the token. They can win by defecting from the contest altogether.
The decisive variable is whether the instrument is new or outstanding. Moving an existing listed share from the account chain to a direct register requires the register-keeper to cooperate and the issuer to want it, security by security, against a settlement system that already works. The SEC's exemption gives an issuer the right to object when a third party tokenises its stock, and the first venue notice filed under it already recorded an objection: a small illustration of the friction that appears the moment issuer consent is asked for. A new bond or fund unit needs no migration: it can be born on the register. The consequence is that the answer differs by instrument, and has to be given instrument by instrument.
Put the same question to each: an entry in what, kept by whom?
Fund units are the cleanest case, and the least noticed. A fund's transfer agent has always kept the constitutive record of unitholders; the account chain was a distribution convenience laid over it. Tokenising a fund therefore changes the medium of a register that was already direct. Franklin Templeton's on-chain money market fund, the exhibit in Section I, states the arrangement precisely: the transfer agent "maintains the official record of share ownership via a proprietary blockchain-integrated system", and the shares on it remain under the transfer agent's unilateral control, including the power to correct entries. The register is the registrar's; the chain is how it is written. Here tokenisation builds on an existing transfer-agent framework; it needed no new statutory category and no depository's consent. This is the instrument where the registrar's book has become the market, and that is one reason tokenised funds grew first.
Bonds go native because each issue is new and the issuer chooses the form. Siemens issued a €60 million bond in 2023 as a crypto security under the eWpG, on a public chain, with no depository in the structure. That is a direct register, applied once. But most of what is called a digital bond does not look like that. Many sovereign and supranational issues are recorded on a platform operated by, or booked into, a depository, and investors hold through its members: a tokenised record at the top of an unchanged account chain. The double application applies to debt as much as to equity. The label "digital bond" says nothing about which pair an issue belongs to; the terms and conditions do.
Listed shares are the hard case, for the reasons in Section IV: the stock is outstanding, the venue rules presuppose the depository, and the issuer must want the change. They stay wrapped, or tokenised in place.
Funds were always on a register. Bonds can be born on one. Shares would have to be moved, and few with the power to move them have yet had a reason to.
One class falls outside the frame, and saying so marks its boundary. A synthetic exposure — a swap, a contract for difference, a perpetual — is a contract with a counterparty or a clearing house. There is no register that constitutes ownership of it, only a position and a margin account. A tokenised synthetic that tracks a share price is therefore not a weaker form of tokenised share; it is a different kind of thing, and the SEC's order draws the same line by excluding synthetic exposure from the exemption. The question "an entry in what?" has no answer there, and the right question is the older one: who owes me, and with what collateral?
The base case, then, is asymmetric. Fund units and new debt are the easy candidates for native issuance, because record and terms can be designed from the start. Outstanding listed equity carries the highest coordination cost and mostly stays wrapped or tokenised in place. This is a forecast about incentives, not a legal necessity, and one large issuer moving its register would be evidence against it.
Who keeps the constitutive register is not being decided by competition. In most places the legislator or the regulator has decided it in advance, and differently each time.
Four answers, none of them market outcomes. A firm planning a single global tokenisation model is planning for a jurisdiction that does not exist.
The record layer is the last that will globalise, if it does at all. A register is constitutive because a national property law says so, and property law is the competence states surrender last. The cautionary tale is the Hague Securities Convention: a conflict-of-laws instrument only, concluded in 2006, asking states merely to agree which law governs an intermediated position. It is in force for the United States, Switzerland and Mauritius. The European Union never signed it — the Convention's free choice of the account agreement's law sat badly with the Union's own place-of-the-relevant-intermediary approach. The uptake proves nothing about the future, but the analogy is hard to avoid: states that would not agree on the pointer are unlikely to agree soon on the thing pointed at.
Standardisation arrives anyway, by a different route. The dollar stablecoin became a world standard without a treaty, because liquidity chose it and every other jurisdiction then had to decide how to receive it. The tokenised US large-cap may be on the same path. If US liquidity and regulatory access outweigh local native alternatives, the instrument the world holds will not be a harmonised native share; it will be an American account-chain position with a tier added, a custodial token that foreign law has to classify on arrival. Where the custodial token cannot be distributed, local issuers can build the wrapper proper on top of it, and the holder is then two structures away from the share. A de facto standard of that kind would be set from above, by the deepest pool, and it would be an account-chain standard. Where the law gives the issuer a say — as the US order does, through its thirty-day right to object — that standard is built share by share, with the register's consent, and the first weeks suggest the consent will not be general.
Between unharmonised registers and a gravitational standard sits the bridge, and the bridge is the prize. Euroclear was founded in 1968 to settle Eurobonds that belonged to no national system, and the position it took, the place where incompatible national records meet, became the most durable franchise in post-trade. The token-era equivalent is whoever maps an entry on one register to an entry on another and is trusted by both. "Interoperability" is the engineering word for it. The commercial word is toll. Note the division this implies: the register holds the legal power, and the bridge may hold the rent. A record can be constitutive within its jurisdiction and still not be the franchise worth owning.
The legal patch is the register's own law. Swiss private international law lets the law designated in the ledger-based security govern it; the UNIDROIT Principles take the same approach, looking first to the law specified in the asset, then in the system. The German rule is different in kind: the governing law follows the state that supervises the register-keeper, which is elective only in the sense that an issuer chooses its registrar. The TVTG applies to tokens issued from Liechtenstein, or where the parties expressly choose it. (PILA Art. 145a; UNIDROIT Principle 5; eWpG § 32; TVTG Art. 3) These rules work inside a single register. They are untested at the seams: when a position converts between an account chain and a register, and in the two-law wrapper of Section III, where each layer answers to a different forum. No convention will settle those seams. Insolvency courts will, one failed tokeniser at a time, and the first few judgments will be the standard.
None of this removes the intermediary. Custody re-intermediates for reasons that have nothing to do with ledgers: licensing, tax reporting and withholding, distribution, and the ordinary preference not to be responsible for a private key. Crypto is the proof. An asset designed for self-custody is held, in the main, at exchanges and custodians, and increasingly inside funds.
What can change is the intermediary's legal position. On a whitelisted direct register the investor is the holder of record and the intermediary operates the wallet. It is demoted from the party at whom the investor's right exists to the party who administers access to a right that exists elsewhere. That is the structural gain: the investor's title depends far less on the intermediary's solvency. It is not the end of dependence. Key management, the registrar's continuity, its powers to freeze and correct, and the law's treatment of the register all remain.
The strongest objection is that this gain has been on offer for decades and very few took it. Direct registration exists in the United States; investors hold in street name regardless, because the account chain delivers netting, margin and securities lending, and the register does not. Tokenisation changes that calculation only where a register entry can do what an account entry does: be pledged, moved and released intraday without leaving the register. So expect a barbell. The mass market stays hosted and largely indifferent. Direct holding appears where it is functionally required, which means collateral.
The prediction that follows is specific. Incumbents will not fight tokens; they are issuing them. They will fight direct registration, and the fight will not look like one, because it is conducted in rulebooks. Four instruments do the work.
None of these requires opposing anything. Each can be defended as prudence, and several are.
Who breaks ranks? Not retail investors, who have no reason to care. The pressure comes from three places. Issuers with a treasury motive: fund managers and repeat bond issuers who save the depository's fee and see their holders. Transfer agents, for whom the direct register is the first growth story in a generation. And collateral takers, which is the one that matters. Once a clearing house or a prime broker accepts a tokenised fund unit as margin directly on its register, with no depository in the chain, routinely and at a discount that competes, the register has done the thing the account chain was needed for. That is the wedge, and it is why tokenised money market funds, not tokenised stocks, are the instrument to watch.
One plausible end of the fight is neither victory nor defeat but occupation. A depository can become the registrar: keep the direct register itself, under the new statutes, and offer it as a service. German law already provides for it: a central register for electronic securities may be kept by the depository, or by a custodian the issuer authorises. Where that model spreads, the title chain shortens in law and the occupant of the seat does not change. Holders would gain what this section began with, a title that depends far less on an intermediary's solvency, and incumbents would keep the franchise. That would be a real improvement and no revolution. Whether it becomes the general pattern or stays a German arrangement depends on the events below, and on issuers whose preferences are not yet visible.
Four observable events would show which way it is going:
The first would mean the register has won the function. The last would mean it has won the seat. Until one of them happens, most tokenisation is a change of medium.
Tokenisation ends the intermediated title chain where three things coincide: a new instrument, a statute that makes the ledger the register, and a use that needs the holder on it. Everywhere else it adds a link to the chain and calls the link ownership. The register stays national, the custodial token is the likeliest candidate for a standard nobody agreed to, and the franchise worth having is the seat between registers.
For any tokenised security, ask the question twice: once of the token, once of what stands behind it. An entry in what, kept by whom, under whose law?
If the two answers differ, there are two instruments, and only one of them is on the chain.
Written in a personal capacity. Analytical views only — not legal or investment advice.
Julian Gretzinger — Investor and writer on monetary history, real wealth mechanics, and financial markets. substack.com/@juliangretzinger