Which blockchains RWAs live on, and why

A map of where RWAs live

Every chain on this map won a specific piece of business, usually because a large issuer arrived there first or because the chain was built for one job.

Two terms from chapter 1 matter for any per-chain number. Distributed value is tokens that can leave the issuing platform and move between wallets. Represented value is tokens that only record a position on a chain and cannot move freely. Rankings by chain are almost always rankings of distributed value.

Ethereum's lead in value is real, and it comes from history more than from current technology. On 3 August 2026 it held 17.12 bn USD, 45.8% of distributed RWA value across 38 chains, on RWA.xyz figures relayed by Coinpaprika. BlackRock, Franklin Templeton and Ondo launched there first, and an issuer that picked Ethereum in 2024 adds chains in 2026 rather than leaving.

Someone often pays for those additions. BUIDL charges 50 bps on Ethereum, Arbitrum and Optimism and 20 bps on Aptos, Avalanche and Polygon, because the foundations behind those three chains pay BlackRock a quarterly fee to cover the difference, as CoinDesk reported on 13 November 2024. A chain foundation is buying a distribution slot.

Growth points elsewhere. In the six months to 26 July 2026, BNB Chain added about 3 bn USD of RWA while Ethereum added about 0.4 bn USD, the smallest gain of the top five. And RWA growth on a chain does not lift its native token: Stellar's RWA value rose about 360% during 2026 while its XLM token fell about 11% over the same period, because institutions can use a chain while holding as little of its token as possible.

Value held versus value traded

Holding value and trading it are two different jobs, and in 2026 different chains did them.

Turnover ratio is transfer volume over a period divided by the value held, which shows how often the average token changes hands. It separates a market from a register. On the XRP Ledger in August 2026, represented RWA turned over about 0.45% a month, while stablecoins on the same ledger turned over about 431% a month. Fees, speed and finality were identical on both sides.

What differs is who may receive the token. Any address can receive a stablecoin. A tokenized fund goes only to an address on a transfer agent's allow-list, and no amount of throughput changes that.

Ethereum keeps the other job. Collateral is an asset pledged to secure a loan, and a tokenized fund becomes useful collateral only where a lending market accepts it. Aave Horizon, Morpho, Pendle and Sky all sit on Ethereum, which is why chapter 8 spends most of its time there. Value parked measures issuer confidence, and volume measures market demand.

XRPL also shows what a headline can hide. "XRP Ledger passes 3 bn USD in tokenized RWA" was true in spring 2026, and about 55% of it was one represented energy token open only to non-US investors. The distributed market excluding stablecoins was about 323 mn USD held by 199 addresses.

What compliance costs in gas

On a public chain, the rules of a regulated token run as code on every transfer, and code costs gas. Gas is the unit Ethereum uses to meter computation, which users pay for in ETH.

I measured the median gas of a plain transfer call on Ethereum mainnet on 4 September 2026. USDC, a stablecoin with a simple design, used 45,160 gas units. BUIDL used 183,672, or 4.07 times as much, and Ondo's OUSG sat between them at about twice USDC.

The reason is visible in the call graph. Every BUIDL transfer asks a compliance service contract to check the investor registry, investor-count limits, accreditation, lock-ups and the mapping of wallets to investors, then updates the records. A dozen calls between contracts replace one storage write. Chapter 4 reads the same contracts for admin powers.

For a fund that trades rarely, four times the gas matters little. For a token meant to move often, it is a tax on every trade, and it is one reason permissioned issuers look at chains where the rules live in the protocol.

Compliance against composability

Chains with compliance built into the protocol give up composability, and chains with the most composability can enforce compliance only through contracts. No chain in 2026 had both.

Composability is the property that lets any protocol on a chain use a token as a building block: a lending market accepts it, a DEX pools it, a vault wraps it. On Ethereum or Solana, a token can reach every venue. On a closed network, it reaches one.

The largest tokenized asset in the world shows the trade. A HELOC, a home equity line of credit, is a revolving loan secured on the equity in a house. Figure Technologies tokenized a portfolio of HELOCs worth about 20.1 bn USD on 7 July 2026, more than all tokenized US Treasuries combined. It lives on Provenance, a chain Figure built for that purpose, where the entire value locked belongs to Figure's own market. Figure's case rests on cheaper loan processing, and the saving it cites is a company claim that no audit has confirmed.

Canton is the institutional version of the same choice. Each participant keeps its own ledger, and a shared synchronization layer lets two ledgers change together or not at all, so a position is visible only to the parties to it. That design suits repo and collateral deals between banks. DTCC and Digital Asset received authorization to tokenize a subset of US Treasuries held at DTC on Canton, with rollout targeted for the second half of 2026. Canton's own volume figures, such as trillions of dollars processed, cannot be verified from outside and are notional flows, which should never be set beside the stock of distributed RWA.

Privacy: deployed and announced

Institutions do not want competitors reading their positions on a public chain, so privacy comes up in every tokenization pitch. Most of what circulates has not shipped.

Two things are deployed and confirmed, and neither hides data with cryptography. Chainlink's Automated Compliance Engine, launched on 30 June 2025 with Apex Group, GLEIF and the ERC3643 Association and adopted by Aave Horizon, answers whether a participant meets a rule without the protocol storing the participant's documents. ERC-3643's ONCHAINID stores signed claims about an investor as references and hashes, never personal data.

A zero-knowledge proof lets one party prove a statement true without revealing the data behind it. Solana's Token-2022 program has a Confidential Transfers extension built on one, which hides amounts but not the parties. When I read BUIDL's Solana mint on 5 September 2026, six extensions were switched on and Confidential Transfers was not among them. Canton's privacy works without any proof system, by design.

Ondo Chain and hardware enclaves for private computation were both announced, and no source establishes their production state. I found no deployed, chain-readable compliance based on zero-knowledge proofs in any of the six largest tokenized funds.

How a regulated token moves between chains

A stablecoin crosses chains through a bridge: lock or burn on one side, mint on the other, and anyone can use it. A permissioned token cannot work that way, because the allow-list must follow the token. A bridge that minted BUIDL to any address would break the rule that only approved investors may hold it.

So the transfer agent does the job. A BENJI holder who wants to move a position coordinates with Franklin Templeton's transfer agent, which burns the tokens on the source chain and mints the same amount on the destination once both addresses are on the allow-list. The official register records one position throughout, and each chain shows its slice.

Ondo moves OUSG between Ethereum, Mantle and Polygon through a bridge that checks the allow-list on both sides, rather than a public bridge. Chainlink's CCIP, a cross-chain messaging service generally available since 24 April 2024, offers the plumbing for this kind of transfer, and its Digital Transfer Agent standard, launched with UBS in September 2025, aims to let transfer agents process subscriptions and redemptions across chains.

This is also why per-chain statistics mislead. BUIDL and BENJI each run on eight or nine chains, depending on the source, and the same share class can grow on one chain and shrink on another without any investor buying or selling.

What to watch next

• Canton's Treasury rollout with DTCC: the first verifiable balances would show how much institutional value sits on a private network, which today is a claim.
• Tokenized-stock volume outside Solana: Robinhood Chain and other layer 2s are the first serious test of Solana's share of trading, covered in chapter 9.
• Foundation-paid share classes: whether more chains pay issuers to deploy, and whether the paid classes attract holders once the payments stop.
• Confidential Transfers in a large fund: the first major fund to switch on Solana's privacy extension would move privacy from roadmap to production.
• Cross-chain transfer agent standards: adoption of burn-and-mint by transfer agents across chains would make multi-chain funds behave like one register.