Arbitrum's Tokenized Asset Market

01 · A market on paper

Everything is here except the owners

The promise is simple. Take a thing that exists off the chain, a share in a firm or a claim on a Treasury bill, and issue a token that stands for it. The token settles in seconds, moves at any hour, and anyone can hold it. The asset stops being trapped inside one broker's systems.

Arbitrum has taken that promise up. It carries share tokens for a few thousand firms, money market funds from BlackRock, Franklin Templeton and WisdomTree, gold, private credit, and dollar tokens that pay a yield. Priced at today's quotes the pile comes to more than five billion dollars, which would make it larger than every stablecoin on the same chain.

That figure needs taking apart before it means anything. Four fifths of it is minted supply sitting in the wallets of the firms that issued it, in tokens that have hardly ever moved. We identified those wallets by what they do rather than by how much they hold: each one received the mints of its issuer's whole book and has sent almost nothing on. Together they account for 81.4% of the total. Set them aside and what anybody else holds comes to $1.01B.

Bitquery indexes every Arbitrum block, so we went through it asset by asset. We found the contracts from their creation records rather than their names, read the supply of each one directly from the chain, priced them, and then asked the question the dollar total does not answer: who holds this. The count that matters is how many sit in the hands of someone other than the firm that issued them.

The answer is that almost none are. The market is real in the sense that the assets exist and the prices are honest. It is a warehouse in the sense that the stock has never left the building.

02 · What is actually here

Four asset classes, and one of them is almost all of it

The market splits cleanly across the real-world asset classes. Share tokens are the bulk of the dollars. Money market and Treasury funds come second, and hold the most real money in the sense that people other than the issuer own them. Dollar tokens paying a yield are third. Gold and private credit are rounding errors.

03 · Four fifths of the total is one wallet

A thousand share tokens, minted, priced, and never sent anywhere

Of the 81.4% held by issuers, one firm is almost all of it: 78.6% of everything on this chain by value, on its own. It has minted 1,020 separate share tokens, covering companies from Apple to a Milanese football holding company, and priced at last close they come to $4.27B. Every one of them sits in a single wallet. Across all 1,020, tokens have left that wallet on fewer than a hundred occasions in the book's life.

The gap between that and any outside measure is wide. DefiLlama, which tracks what these tokens are worth in circulation rather than what has been minted, records $1,045 of this issuer's assets on Arbitrum and $3.2M across every chain it works on. Press reporting this month put the firm's assets under management near eight hundred million dollars, still a fraction of what has been minted here.

Those numbers do not contradict each other, because they count different things. Minted supply is what a contract says exists. Circulating value is what somebody other than the issuer holds. Here the first is $4.27B and the second rounds to nothing, and only the second is a market. We cannot see the issuer's books, and no ledger can say what backs a token. What a ledger can say is where the tokens are, and they are all in one place.

04 · Who owns any of it

Forty-four thousand addresses, two thousand positions worth more than a hundred dollars

Counting holders on a chain is easy to get wrong. A token will happily report a wallet as a holder when it sits on a fraction of a cent, and airdrop spam pushes those counts into the tens of thousands. So we took every wallet that has ever received one of these tokens, asked each token directly what that wallet holds today, and then priced the answer.

Forty-four thousand wallets hold a balance. The median position among them is worth less than a tenth of a cent. Raise the bar to a single dollar and the count falls by five sixths. Raise it to a hundred dollars, a small trade by any standard, and two thousand wallets remain on a chain that sees millions.

Before reading that as indifference, note who is allowed to buy. Several of these products are sold only to qualified or professional investors, and one large book of share tokens is offered to European clients rather than American ones. A short list of owners is partly a fact about permission.

Size predicts whether a holder stays. Split every wallet by how much it held on the day it arrived and the pattern is steep: of those that started with a real position, four in ten came back within the month and one in nine was still active a quarter later. Of those that arrived with a token or two, one in eight came back and one in sixty was still there. The market has a small core that persists and a large fringe that passes through.

Three wallets explain the gap between the dollar total and the owner count. One holds every share of 821 stock tokens, worth about four billion dollars, and has passed a token to somebody else 98 times in its life. A second holds all 1,616 of another issuer's tokens. A third holds all 624 of a third's. None of them is doing anything improper. They are what a broker's own book looks like once it is written to a public chain.

05 · More tickers, fewer people

The market is growing on the supply side and shrinking on the demand side

Six months of this market against the six before gives a clean read. The count of tokens that have issued anything roughly doubled. The count of wallets that touched one fell by three quarters, and the count doing it for the first time fell by more than nine tenths.

Part of that fall is a wave washing out rather than a crowd leaving. One month in late 2025 brought 47,111 first-time wallets into the fund layer, several times any other month. Ten per cent of them ever came back, and 1.3% still hold anything today. Strip that month out and the trend is the same, only gentler.

Retention across every cohort since the start of 2025 tells the same story. Of 101,132 wallets that received one of these tokens for the first time, 19.2% came back after a month and 3.4% were still active in the last quarter.

06 · Supply grew hard, then reversed

Four hundred million added in half a year, then fifty million taken back

Holder counts describe attention. Supply describes size. For the funds that mint and redeem on Arbitrum itself, every issue and every burn is written to the chain, so the change in size can be read directly rather than inferred. What that change means takes one more step, and the step matters.

The direction changed this year. The six months to February added a net $436.4M of supply. The six months since have taken $54.5M back out. That is a swing of roughly half a billion dollars in a market whose whole fund layer is smaller than that.

What that swing is not is a crowd of investors heading for the door. Look at who does the burning and 71.4% of it by value is smart contracts rather than wallets. The three largest are a protocol proxy, a swap adapter, and a cross-chain pool that burns a token here so it can be minted somewhere else. At least $140.2M of the outflow is tokens changing chain, which is a different event from an investor being paid out. The honest reading is that supply on this chain grew hard and then shrank, and that a good part of the shrinking is stock moving to another shelf.

One thing it is definitely not is the chain's own treasury leaving. Arbitrum's DAO and Foundation wallets account for none of the redemptions in this window at all.

07 · What the funds actually pay

Two government money funds, thirty-seven basis points apart

Some funds pay their yield the way a bank credits interest, by issuing more units to holders. Where that happens the rate is readable: take a day's credit, divide by the supply it was paid on, and the annual figure falls out without anybody having to publish a factsheet.

It happens rarely. Of 17 funds here, 2 pay that way. The rest either accrue in the price of the share, so nothing is minted and the chain shows nothing, or mint only when somebody subscribes. For those, a reader gets no more from the chain than from a printed brochure, and a rate read off their mints would be nonsense: one of them implies five hundred per cent a year, because subscriptions are being counted as dividends.

The two that do publish run on visibly different plumbing. One credits every holder, every calendar day including weekends. The other pays on business days only, routes the whole fund's accrual through two addresses, and settles the weekend on Monday.

08 · Does any of it move

The funds circulate. The shares do not.

A transfer is not evidence of a market. An issuer minting tokens to itself and burning them again fills a ledger all day without anybody trading anything. So we split every transfer in two: the issuer making or unmaking supply, and one wallet sending to another.

That split is the sharpest number here. In the fund layer, almost all movement is wallet to wallet. In the share layer, three quarters of it is the issuer minting and burning against itself.

09 · Whether it can move

Most of this market is not allowed to trade

Before reading the quiet as a lack of demand, it is worth asking whether trading is allowed at all. We asked each token directly, by trying a send to a wallet that is not on the issuer's list and noting what came back.

The funds refuse, and say so in words their issuers wrote: a wallet not in the registry service, a shareholder that does not exist. Thirteen of the nineteen behave that way, and they hold 63.6% of the money in the fund layer. An exchange pool is a wallet like any other, so a token that will not send to a stranger cannot sit in one. For the funds, the quiet on exchanges is a design choice rather than a verdict from buyers.

The share tokens are a different case, and the answer surprised us. We ran the same test across 3,629 of them and 3,626 let the transfer through. Three refused. These tokens are free to move, free to be sold, free to sit in a pool. Nothing in the code stops any of it.

So the emptiness of the share layer needs a different explanation from the emptiness of the fund layer. The funds are restricted and held by a handful of approved investors. The share tokens are unrestricted and held by nobody.

10 · How Arbitrum compares

Fourth place, and most of its total is two assets

Arbitrum is not where this market lives. Take the funds that exist on several chains at once and compare where their supply sits: Arbitrum holds about a ninth of the total and Ethereum about half. Avalanche and Stellar are both bigger, each on the strength of a single fund. Arbitrum's own share is two assets carrying almost all of it.

11 · How much of the chain this is

One transaction in 461

Set the tokenized market against the chain that carries it. Over ninety days, about one Arbitrum transaction in 461 touched a tokenized real-world asset. USDC alone clears more transfers in four hours than the entire non-stablecoin tokenized market produces in a month.

That is the honest frame for the five billion dollar total. The dollars are real and they are large. The use sitting on top of them is a rounding error on a busy chain, and the gap between the two is what this market has to close.

We keep finding the same shape. The nine-chain census put most of the value in a handful of tokens, the xStocks book on Solana put most of the trading in a handful of names, the bStocks book on BNB Chain turned out to trade while New York slept, and Robinhood's own chain showed the same issuer-side pattern. Arbitrum is the largest version of it by dollar value, and it is worth watching because that is where the promise gets tested.

12 · The record

Addresses and method