Tokenized Stocks Are Supposed to Be Collateral. Almost None of Them Are.

An idle share earns nothing.

That sounds too obvious to say. It is also the real economic argument for putting equities on a blockchain and almost nobody makes it out loud. The pitch you hear is about trading hours. Buy Apple at 3 a.m. Buy it on a Sunday. But hours are a feature, and features get copied. NYSE has a 24/7 plan of its own, and when it ships, the headline reason to hold a token instead of a share evaporates.

Collateral is different. Collateral is a business.

A share posted as collateral does two jobs at once

Here is the mechanic that matters. You own Apple. You think it goes up. If you sell it to raise cash, you lose the position. If you pledge it instead, you keep the exposure and you get the cash and whoever lent you the cash charges you for the privilege. One asset, two jobs.

Traditional finance has run this business for a century. It is called securities lending and margin financing, and it is enormous. The Depository Trust Company, the entity that actually holds most American stock, crossed $100.3 trillion in assets under custody in June 2025, of which $74.1 trillion was equities. A large slice of that is pledged or financed at any given moment. Aave, announcing its plans in June 2026, framed its own target as a $4.6 trillion securities lending market, which is Aave’s number and worth treating as an ambition rather than a measurement.

So the bet behind tokenized equity was never about hours. The bet was that a share on a public blockchain becomes composable, meaning any program can hold it, price it, lend against it and liquidate it without asking permission from a broker. Post AAPLx into a lending pool at 2 a.m. on a Saturday and borrow stablecoins against it and nobody signs anything.

That is a genuinely new capability. Which makes the next number strange.

The collateral use case is live, and it is tiny

Tokenized-stock lending TVL, meaning the total value of tokenized shares posted as collateral across DeFi lending markets, was $23.1 million on 17 July 2026, per Token Terminal data. Solana held 85.5% of it. One protocol, Kamino Finance, held 82.6% of the Solana share, so roughly $16 million sat in a single venue.

Set that against supply. Tokenized equity on Solana alone hit $535 million on 16 July 2026, an all-time high. The whole category was $2.3 billion in July 2026 and about $2.5 billion by mid-August. So somewhere around one percent of tokenized stock was doing the second job. The other 99% was sitting in wallets, doing what a share in a brokerage account does: nothing.

This is not unique to equities. DefiLlama data cited in early September 2026 put the whole tokenized real world asset market at $34.6 billion with only $3.79 billion deployed into protocols, roughly 89% idle. Tokenized stocks are worse than that average, not better.

Kamino was first, in July 2025, starting with a single ticker. Here is the map as of now.

The tokenized-equity collateral map, September 2026

Two rows in that table are the article. Aave is the largest lending protocol in crypto and it announced tokenized stock lending with real fanfare. As of 30 August 2026, an inspection of its Base deployment found no reserve, no aToken, no debt token and no oracle entry for Coinbase’s tokenized stocks. The integration is a stated goal. The market does not exist yet.

And the DTC row is stranger still. I will come back to it.

What the leverage stack looks like when you count the layers

Kamino does not stop at plain borrowing. Its Multiply product loops a position in one atomic transaction: flash borrow USDC, swap it into SPYx, deposit that as collateral, borrow against it to repay the flash loan. Repeat the loop and you hold more S&P exposure than you paid for. xStocks has promoted this directly, describing users looping SPYx, QQQx and TSLAx to lever up their exposure onchain.

Now count what is actually stacked underneath a leveraged SPYx position.

1. Real S&P 500 shares, sitting with a regulated custodian off-chain.
2. A security entitlement against the entity that holds them.
3. A token representing that entitlement, which is a claim on a claim, not a share.
4. A lending market that accepts the token, priced by an oracle rather than by the exchange where the shares trade.
5. A stablecoin loan against it.
6. More of the same token, bought with that loan, pledged back into layer four.

Every one of those layers is someone’s promise. The blockchain verifies layer three onward and can say nothing at all about layers one and two, which is the whole design of the custody chain.

Rehypothecation, meaning the reuse of collateral you are holding for someone else, is the traditional word for what layer six does. On-chain it is not quite rehypothecation in the legal sense, because the protocol is not relending your specific deposit to a third party by contract. Functionally the effect rhymes. The same underlying share is supporting more than one position and the number of claims that ultimately reference one custodied share grows with the size of the loop.

Then add the timing problem. Liquidation logic runs continuously. The exchange that produces the real price does not. A leveraged position in tokenized equity can be closed out at a Saturday price nobody else in the world is trading at and no better oracle fixes that.

I want to be precise about the stakes here, because the scary version of this section writes itself. At $23 million, nothing about this threatens anything. What has been built is a small working prototype of a mechanism that would matter enormously at scale. The machine runs. Nobody is feeding it yet.

What the SEC actually said on 28 January 2026

On that date, staff from three SEC divisions, Corporation Finance, Investment Management and Trading & Markets, published a joint statement on tokenized securities. It is the most consequential document in this market and it is routinely described in ways it does not support.

Start with what it is. In the statement’s own words: “This statement represents the views of the staff… It is not a rule, regulation, guidance, or statement of the U.S. Securities and Exchange Commission,” and it “has no legal force or effect: it does not alter or amend applicable law.” It is staff telling you how they read existing law. It bans nothing.

What it does is build a taxonomy and the taxonomy is the point. The staff define a tokenized security as “a financial instrument enumerated in the definition of ‘security’ under the federal securities laws that is formatted as or represented by a crypto asset.” Then they split the universe in two: securities tokenized by or for the issuer and securities tokenized by third parties unaffiliated with the issuer. Third party tokens split again, into custodial and synthetic.

For custodial tokens, the staff describe a third party creating a security entitlement formatted as a crypto asset, where the token “represents the holder’s indirect interest in the underlying security via the security entitlement.” Their working assumption is that this does not create a new security.

For synthetic tokens, giving exposure without conveying ownership, the assumption flips. Those may be a separate security and may be security-based swaps. The operative sentence is blunt: a third party “may not offer or sell the crypto asset representing the security-based swap to persons who are not eligible contract participants unless a Securities Act registration statement is in effect… and the transactions… are effected on a national securities exchange.” An eligible contract participant is, roughly, an institution or a wealthy individual. Retail is not one.

Statement versus inference

Keep these apart. Commentary has blurred them badly.

The right hand column may well be correct. It is not what the document says and the difference matters if you are building something.

The Empty Quadrant

Here is the frame I keep coming back to. Draw two axes.

The vertical axis is legal realness: how close is this token to being the share itself, with the same rights and the same recourse? The horizontal axis is composability, meaning how freely can any program hold, price, move and pledge it without permission?

Plot what exists.

Bottom right, real and composable but legally thin: xStocks on Solana. A custodial entitlement, freely transferable, usable in Kamino and the reason 82.6% of tokenized-stock lending happens in one place. Also non US only.

Middle right: Coinbase’s B20 tokens, launched on Base on 24 August 2026 under Abu Dhabi Global Market regulation and Regulation S. No whitelist once minted, so genuinely composable, but the issuer can still freeze wallets and US persons are excluded.

Top left, legally real and barely composable: the DTC pilot tokens. The underlying stays registered to Cede & Co. Transfers happen only between DTC participants, in registered wallets.

Top right: empty.

Nothing today is both fully the share and fully permissionless. That is not a temporary engineering gap. The January statement priced the trade-off: the further a token moves from the issuer, the more likely staff treat it as its own instrument with its own registration problem and the closer it stays, the more of the issuer’s transfer machinery it drags along.

Composability and legal realness pull against each other. Every product in this market is a choice about which one to give up.

Read the DTC pilot’s fine print

Which brings me back to that row in the table and to the single most underreported fact in tokenized equity.

On 11 December 2025, SEC staff granted DTC no-action relief for a three-year tokenization pilot covering Russell 1000 stocks, major index ETFs, and Treasuries. Participants can tokenize security entitlements into registered wallets. Transfers between those wallets can happen 24/7. Real progress.

And then, in the conditions: tokens do not receive any collateral or settlement value for DTC risk management purposes.

Read that again. The incumbent built the legally real tokenized share, the one thing crypto issuers cannot build and switched the collateral function off. Deliberately, as a risk management condition of getting the relief at all.

Nasdaq’s side moved on the same pattern. On 18 March 2026 the SEC approved SR-NASDAQ-2025–072, letting eligible participants trade tokenized versions of liquid equities and ETFs on the same order book with the same execution priority as traditional shares, provided the tokens are fungible with and afford the same rights as the underlying security. Members get at least 30 days’ notice; earliest live trades were expected in Q3 2026 pending DTC onboarding. Then on 9 March 2026 Nasdaq announced an equity token design built around issuers, aimed at corporate actions, proxy voting and governance, with a target of going live in H1 2027.

Notice what that program is optimizing for. Proxy voting. Corporate actions. Shareholder engagement. Faster settlement. Not collateral, not composability, not a program you don’t control holding your stock.

Who structurally wins

I think the incumbents win the collateral business and I think they win it slowly and boringly. Collateral is a credit and legal business wearing a technology costume. A lender needs three things: an enforceable claim, a price it trusts and a way to seize the asset when things break. Blockchains improve the last one. They do nothing for the first and the four issuers are four different legal products precisely because the first thing is hard.

DTC starts with $74.1 trillion of equities and every legal relationship already papered. Getting from there to a token is an infrastructure project. Getting from a Jersey issued bearer note to $74 trillion of enforceable claims is not a project, it is a decade.

But I should say what would prove me wrong, because the case has a hole in it. The incumbents’ token is currently worth zero as collateral by rule, moves only between DTC participants and is aimed at proxy voting. If that stays true, the incumbent “win” is just faster settlement of the existing system. Useful. Not the bet anyone was making. Meanwhile the composable version works today, has a live leverage product and Ondo Perps proved in July 2026 that people will post tokenized shares as margin when you let them, with up to 20x leverage on Apple and Nvidia, for everyone outside the United States.

So the realistic outcome is two systems that do not touch. A regulated one with all the assets and none of the composability and an offshore one with all the composability and about one percent utilization. The interesting question is not who builds better software. It is who gets permission to connect them, and that is a legal question with a legal answer, arriving on a legal timetable.

What the last four articles were building toward

The series has been one argument told in five parts. A tokenized stock is a claim on a claim. That claim depends on one boring off-chain custody step. Four issuers turned that step into four legally different products. Trading them around the clock means trading them at hours when nothing can tell you what they are worth.

All of that is overhead. You accept every bit of it for one reason: because the thing you end up holding can be pledged, looped and put to work by a program at three in the morning without asking anyone. That is the payoff the whole structure exists to deliver.

Right now, ninety-nine cents of every tokenized dollar is not collecting it.

An idle share earns nothing. Turns out an idle token doesn’t either.

This is part 5 of a five-part series on tokenized equities. Nothing here is investment advice. All figures are current as of September 2026; this market moves fast, and some of these numbers will be wrong by the time you read them.