Tokenised treasuries: what is actually being tokenised

Of everything filed under real-world assets, tokenised treasuries are the part that worked. The product is unglamorous by design: a fund holds short-dated government debt, and ownership of shares in it is recorded on a public chain instead of a private register.

That is the whole innovation, and it is worth stating plainly, because the category attracts explanations that make it sound like something else.

What is being tokenised, exactly?

Not the bond. The share in the fund that holds the bond. The distinction decides everything downstream: the token is a claim on a fund, governed by the fund's documents, subject to its redemption terms, and dependent on its administrator.

A holder of the token owns a fund share whose transfer happens to be recorded on a chain. They do not own the underlying security, and they cannot present the token to a government for repayment.

What the recording change buys is real, though smaller than the marketing suggests: settlement in minutes rather than a day or more, transferability outside market hours, and the ability to use the position as collateral in systems that can read a chain.

Where does the yield come from?

From the government debt, less the fund's fees. That is the entire mechanism, and it produces the single most useful test for anything in this category: compare the advertised yield to short-dated government paper of the same maturity. A product paying materially more is not tokenising treasuries alone.

The excess is compensation for something, and the something is usually one of a short list — credit risk from an adjacent asset, leverage inside the structure, a subsidy from a token, or a promotional rate that stops. None of those is disqualifying, and all of them change what the reader is buying.

The desk's rule when reading any on-chain yield: identify whose obligation the payment is. If the answer is a government, the yield is a government's yield. If the answer is a protocol, the yield is that protocol's credit, whatever the underlying is called.

Who actually holds these, and did that change?

The category passed twelve billion dollars, and the more interesting change was in who owns it. The early holder base was crypto-native: treasuries of protocols and funds parking idle stablecoin balances somewhere that paid the risk-free rate instead of nothing.

The shift has been towards holders who want the position as collateral rather than as savings. That is a different demand, and it has different consequences: collateral has to be liquid at the moment it is called, and the moment it is called is precisely when everyone else is calling too.

A second route into the same product appeared at the consumer end, where a treasury fund sits behind an exchange's earn tab. The user sees a rate; the structure underneath is a regulated fund. Whether that is clear to the person clicking is a disclosure question that has not been settled.

How does this differ from a stablecoin?

The two are frequently discussed together and behave differently in the way that matters — under stress.

The row to sit with is redemption. A stablecoin promises par on demand and is judged on whether it holds. A fund share promises net asset value in a redemption window and is judged on whether the window is open when you need it.

Where does it break?

Redemption under stress

The token moves in minutes; the fund redeems on its own schedule. In calm conditions the secondary market covers the gap. In a rush, the on-chain price is what someone will pay now and the fund's price is net asset value later, and those are not the same number.

Two registers, one owner

Legal ownership rests with whatever the fund's documents say it rests with. When a chain state and a transfer agent's register disagree — after a compromised key, a forced transfer, or a court order — the question of which one governs is answered by law, not by consensus. Every serious issuer has a documented answer, and it is worth reading before it matters.

Servicing, which is where the adjacent category failed

Tokenised private credit reached the same problem first and harder. Issuing a token against a loan is straightforward. Servicing it — collecting, chasing, restructuring, reporting a default — is human work that no ledger performs, and the tokenised version inherits every operational weakness of the original while adding a holder base that expects on-chain settlement speeds.

Treasuries avoid most of that because the underlying services itself. It is the reason this corner of real-world assets worked while others stalled, and the reason to be sceptical when the same infrastructure is pointed at an asset that requires a servicer.

What to check before holding one

• What is in the fund, and at what maturity. Compare the yield to that maturity, not to a stablecoin.
• Redemption terms in writing: windows, notice periods, gating provisions, and who may suspend.
• Which register governs ownership if the chain and the transfer agent disagree.
• Who the administrator and custodian are, and whether either is affiliated with the issuer.
• Transfer restrictions. Many of these instruments are permissioned, and a token that cannot move to your counterparty is not collateral.
• What happens to accrued yield on transfer — it is handled differently across products and is a routine source of surprise.

Where this is going

The direction of travel is towards these instruments being used as collateral rather than held as savings, which puts weight on exactly the mechanism that has not been tested at scale: redemption in a rush. Nothing about the structure prevents it working. Nothing about it has demonstrated that it does.

The second direction is regulatory. Licensing dates for stablecoin issuance are now on calendars, and the same supervisory attention lands on funds distributing shares through public chains. That is a cost and also the thing that makes the category investable for institutions that cannot hold an unregistered claim.