How a security lives before anyone tokenizes it

Tokenization did not arrive in an empty room

The story in which tokenization repairs a broken system needs one assumption: that before it there was no system. There was, and it is the one tokenization plugged into.

Book-entry ownership is a security recorded as an entry in a register, with no paper certificate anywhere, and it has been standard practice for decades. A Form N-1A prospectus filed with the SEC in 2025 shows how little the arrival of a blockchain changed about that arrangement. The investor picks whether the shares will be recorded as Book Entry Shares or as Tokenized Shares of the same fund, which makes the chain one of two accepted forms of a single register rather than a separate product.

The structures that hold those registers run at a scale the tokenized market does not approach. Over 2,000 investment funds are registered in the British Virgin Islands, and BVI Business Companies account for over 40% of the offshore entities used in global investment structures, on two 2026 surveys of the jurisdiction. BUIDL's own fund company is one entry in that count, not an exception to it.

So tokenization connected itself to a working and expensive stack, and took one layer out of it. Everything above that layer and everything below it stayed in place, together with its prices. The rest of this chapter is about which layer moved and what the others cost.

Four roles, one job each

Four jobs keep a fund share alive, and every one of them predates the technology in this series. The first thing to establish about a tokenized product is that it still has all four.

The issuer is the party that sells the security and carries the obligation written into the offering documents, as chapter 1 defined it. For BUIDL that party is a company registered in the British Virgin Islands, and BlackRock is the manager rather than the debtor.

A custodian is the regulated firm that holds the underlying asset and may not treat it as its own, which for BUIDL is BNY Mellon. That firm also does the second job.

A fund administrator is the firm that computes the fund's net asset value and keeps its accounting records. Securitize's Form 424B3, filed in 2026, states that the NAV its oracle providers carry to smart contracts is supplied by the relevant fund administrator or transfer agent, which puts the calculation firmly off the chain no matter how the number arrives.

A transfer agent keeps the official list of who owns the shares and processes subscriptions, redemptions and distributions, introduced in chapter 1 and priced later in this one. In the United States it is a licensed role registered with the SEC, so it is a legal function rather than a piece of software, and that distinction decides who can enter this business at all.

Securitize holds four separate regulatory statuses across four entities, as its own July 2026 release sets out: Securitize Markets, LLC as an SEC-registered broker-dealer running a regulated alternative trading system, Securitize Transfer Agent, LLC as the registered transfer agent, Securitize Capital LLC as the adviser entity, and Securitize Fund Services, LLC for fund administration. None of the four can be bought quickly, and together they are what a competitor would have to reproduce before the code mattered at all.

Fund, SPV, feeder and trust: what the holder actually owns

The wrapper is the entity that owns the asset and issues rights against it, and it decides who the holder's debtor is. Four of them carry almost everything in this market.

A corporate fund is a company whose shares are the units of participation, so the holder owns equity in it. BUIDL's fund company is a BVI company limited by shares, registered on 17 September 2023 under number 2132429, with BlackRock Financial Management, Inc. as manager and Securitize Markets, LLC as placement agent.

An SPV is a company whose activity is contractually limited to holding one asset or one pool and issuing rights against it, with no other business and no other creditors. Ondo's documentation for Ondo USDY LLC lists a closed set of five permitted activities: borrowing from lenders, issuing USDY as evidence of those debts, allocating the proceeds to US Treasury bills and demand deposits, granting and perfecting security over its assets, and acts incidental to those.

A feeder fund is a fund whose only asset is shares in another fund, so the issuer tokenizes the feeder and leaves the target fund's prospectus and investor base untouched. Apollo Diversified Credit Securitize Fund, announced on 30 January 2025, is a feeder into Apollo Diversified Credit Fund, and it carries the target fund's terms with it. Both that product and Hamilton Lane's equivalent keep a quarterly redemption gate at 5% of fund assets, processed pro rata, exactly as a traditional non-traded BDC does. The token moves at any hour of any day, and the right to exit stays quarterly and capped.

A trust is an arrangement in which a trustee holds the asset on behalf of beneficiaries, and trust assets are not the trustee's assets. Paxos Trust Company has held a New York limited purpose trust charter since 2015, and its own materials describe an insolvency in which the NYDFS superintendent would return custodied assets to their owners according to the company's books rather than running a standard bankruptcy. This is the only one of the four wrappers where the holder does not have to rely on a creditor ranking, because the asset was never the issuer's property to begin with.

Digital custody costs ten to sixty times the custody it replaces

Two abbreviations run through the rest of this chapter. AUM, assets under management, is the value of what a fund holds and the base its management fee is charged on. bps, a basis point, is one hundredth of a percentage point, so BUIDL's management fee of 50 bps on Ethereum is 0.50% a year, against 20 bps on the Aptos, Avalanche and Polygon share classes announced on 13 November 2024.

Custody in a tokenized product splits into two separate contracts, and conflating them is the most common error in this layer. Custody of the underlying asset answers who holds the Treasury bills, and in BUIDL that is BNY Mellon. Custody of the token answers who holds the private key to the wallet, and the named firms doing that work are Anchorage Digital Bank, BitGo Trust, Coinbase Custody, Fireblocks, Komainu and Copper. Two contracts, two fee schedules, two failure modes.

AUC, assets under custody, is the base the custodian charges on, and the two halves of this layer charge on it at prices that do not belong to the same market. A real custody agreement, the Safekeeping Charges clause of the Sound Shore Fund, runs from 1.00 bps a year down to 0.25 bps as assets grow. Digital custody sits at 10 to 60 bps of AUC a year on 2026 market rates, which is ten to sixty times the traditional price for the same asset. The reason is amortization rather than complexity: a custodian bank spreads its infrastructure over trillions of dollars and several decades, a digital custodian over tens of billions and a few years, and insurance that a bank never prices separately is a line item here.

That assumption decides the shape of the whole budget. In a tokenized Treasury fund built to a 100 mn USD target AUM, digital custody is 250,000 USD of about 435,000 USD in annual running cost, or 57% of it. The largest single line in the budget of a tokenized fund is the cost of holding a private key, and it also explains why BUIDL's minimum subscription is measured in millions rather than hundreds of dollars: under a pricing model charged on assets under custody, a small holder is a loss to the issuer.

The register is the one layer with an arithmetic advantage

Classic transfer agency is priced per account, and the clause that shows it best is old. The IDEX II Series Fund agreement of 1994 charges 12.00 USD per open account per year, 2.43 USD to set an account up and 1.45 USD per closed account, against a contractual minimum of 25,000 USD a year per fund. A fund with 100,000 retail investors pays roughly 1.2 mn USD a year for the register alone, and it pays that every year whether or not anyone trades.

A digital transfer agent is the same licensed role run as a smart contract, and it is priced flat rather than per account, because the hundredth holder and the hundred-thousandth cost the same to add. That is a change of order of magnitude rather than a percentage saving, and it is the one place in the stack where the arithmetic favours tokenization without argument. The price itself is not solid enough to quote in a sentence, so it sits in the table below with its mark.

The advantage has a cost on the other side of the same layer, because a permissioned register checks every transfer and the check is paid for in gas. Measuring an identical transfer call on Ethereum mainnet on 4 September 2026, the author's readings put BUIDL at four times the gas of a plain USDC transfer and OUSG at twice. The same compliance logic that makes the on-chain register legally usable is what makes each move through it more expensive than a transfer with no rules attached.

FOBXX is the longest-running test of this idea. Live since 6 April 2021, it is the first SEC-registered fund to use a public blockchain as a share registry, and its Form 485BPOS describes a hybrid: book-entry record keeping alongside public chains, under the transfer agent's control, rather than a design in which the token simply is the register. Chapter 1 sets out why that distinction decides what a holder owns, and chapter 5 reads the administrative powers a transfer agent needs in the contract to do this job at all.

The correction that follows from all of this is narrow. Tokenization lowers the cost of servicing a security after issuance, in the one place described above, and leaves the cost of issuing it exactly where it sat before. The rest of the promised saving is not visible in what an investor pays.

Break-even is why this market has no small products

Break-even AUM is the fixed annual cost divided by the net fee rate, and that single line decides which products can exist. At about 435,000 USD of annual running cost, a fund charging 50 bps as BUIDL does on Ethereum breaks even at roughly 87 mn USD of AUM, and a fund charging 15 bps as OUSG and USTB do breaks even at roughly 290 mn USD. The same 435,000 USD sits behind both.

Getting to that running cost costs more. The first-year bill for a realistic tokenized Treasury fund aiming at 100 mn USD is about 595,000 USD, which is 59 bps in year one and 44 bps after it, charged before the issuer earns a dollar.

Almost none of that shrinks for a smaller fund. The fixed block covering the legal structure, the issuer entity, compliance and the platform costs a fund of 10 mn USD nearly what it costs a fund of 3 bn, because the offering documents, the contract audit and the entity registration are the same work done by the same people. The one line in that block that a traditional fund does not carry is the smart-contract audit, which adds rather than replaces.

BUIDL shows the same arithmetic from the other end. Its compliance contract, read on-chain on 4 September 2026, caps the fund at 1,999 investors and reported 89 of them in place, 4.5% of the ceiling. A product designed for a few dozen very large holders is what this cost structure produces, and chapter 4 takes the same stack apart company by company.