4.2 Blockchains of Choice and Why They Were Chosen
Where tokenized real-world assets actually live as of a stated date, why the issuers profiled in Section 4.1 chose Ethereum, Solana, Stellar, the EVM layer 2s, BNB Chain, or a permissioned ledger, what regulators and chain foundations decided for them, why almost every product ended up on several chains, and how to read any chain choice.
Section 1.2 makes a claim this section takes as settled: "A tokenized Treasury on Ethereum and a tokenized Treasury on Solana are the same kind of RWA. They differ in settlement properties, not in what they represent." Section 2.3 orders the decisions that produce that sameness: the first decision is not which blockchain to use, it is what right the holder will receive. The chain is the second decision, and it is what this section tracks.
That second decision settles four practical questions. Who can reach the token on day one: which custodians, wallets, venues, and investors already operate there. Who can see each transfer. Where the compliance logic lives: in the token contract itself, in the chain's own token program or asset flags, or in the network's admission rules. And what it costs the issuer and the holder: fees, integrations, and, for a bank, the capital treatment of the ledger itself.
This section documents a pattern rather than argues for one chain over another. Asset managers seeking crypto-native and DeFi distribution went to public chains, Ethereum first. Banks moving their own balance sheets went to permissioned ledgers whose volumes dwarf the public-chain products the same banks also run. One large originator, Figure, built its own chain rather than renting anyone else's. And almost every product that started on one chain ended up on several, in a number of cases because a chain foundation or a DAO paid for the additional deployment.
The data basis for what follows is rwa.xyz's network ranking on 20 September 2026, read through the distributed-and-represented distinction Section 2.3 introduced: whether a token can leave the platform that issued it, or whether the chain serves only as a recordkeeping layer behind an unchanged register. This section takes no position on which chain any issuer should use; it reports which ones they did use, and why.
Where the Value Sits
On 20 September 2026, rwa.xyz tracked 39 networks, with a Distributed Asset Value of $38.53 billion and a Represented Asset Value of $367.99 billion, in the tracker's own labels1. The represented figure is more than nine times the distributed one, and about 95% of it sits on two purpose-built ledgers.
Canton carries $327.02 billion, entirely represented, on a single listed platform: Broadridge's Distributed Ledger Repo2.
Provenance carries $22.76 billion represented against only $83.69 million distributed, all of it Figure's3.
The table below orders networks by distributed value on 20 September 2026, with Provenance and Canton added as the two represented outliers1.
| Network | Distributed value | Represented value | Assets | Holders |
|---|---|---|---|---|
| Ethereum | $16.69B | $79.91M | 2,564 | 273,597 |
| BNB Chain | $5.74B | $0.00 | 2,252 | 1,795,944 |
| Solana | $4.34B | $125.86M | 2,799 | 634,703 |
| Stellar | $3.32B | $78.20M | 72 | 21,044 |
| Avalanche | $1.68B | $11.40B | 1,252 | 10,058 |
| Liquid Network | $1.55B | $0.00 | 9 | 62 |
| Arbitrum | $966.0M | $26.72M | 5,761 | 12,864 |
| ZKsync Era | $962.4M | $1.98B | 50 | 264 |
| Polygon | $506.3M | $727.98M | 405 | 22,935 |
| XRP Ledger | $456.6M | $4.05B | 377 | 4,138 |
| Base | $373.4M | $23.89M | 1,003 | 65,513 |
| Provenance | $83.69M | $22.76B | 2 | 190 |
| Canton | $0.00 | $327.02B | 2 | 0 |
Four readings follow, all on the same 20 September 2026 date.
Distributed value is mostly an Ethereum-plus-EVM market. Ethereum alone carries about 43% of distributed value, with 273,597 holders1.
Solana, Stellar, Liquid Network, and the XRP Ledger are the non-EVM chains that still clear the top ten by distributed value1. Solana, Stellar, and the XRP Ledger are each carried by named products below; Liquid Network's single-platform case appears in the next reading.
Holders and value diverge sharply. BNB Chain's 1.8 million holders sit on $5.74 billion, while Ethereum's 273,597 holders sit on $16.69 billion: the retail pattern the next profile explains1.
The "Robinhood" network row takes the same pattern one step further, 1,398,259 holders on $195.5 million1.
Stellar shows the opposite shape: 72 assets carrying $3.32 billion, few issuers and large tickets rather than many small ones1.
Single products move rankings by themselves. Liquid Network's $1.55 billion is nine assets from one platform, STOKR. ZKsync Era's $962 million is one Securitize asset. BUIDL's Ethereum share class fell from 599.3 million units on 16 September 2026, Section 4.1's date, to 289.4 million units on 20 September 2026, this section's date4. A network's rank can be one product's decision. The move is stated here without attributing it to demand.
What the table omits matters too: rwa.xyz ranks what it tracks, and permissioned networks report what their operators choose to disclose. The Canton figure above is a single platform's self-reported repo book, the same ledger Section 4.1 already named as the largest Represented platform anywhere. That caveat is what keeps this table honest.
Public Chains: What Each One Sells
Public chains sell reach: the custodians, wallets, venues, and investors already present. The price of that reach is transparency of every transfer and, for a regulated balance sheet, the capital and permissioning questions the next two sections take up. Each profile below covers which named products chose the chain and when, what the chain offers at the protocol level for compliance, the dated rwa.xyz figures, and the issuer's own stated reason where one exists, attributed rather than asserted as a finding.
Ethereum: Where the Buyers, the Custodians, and the Standards Are
BUIDL launched on Ethereum on 20 March 2024, already covered in Section 4.1. Fidelity chose the same chain for a narrower reason in 2025: the OnChain share class of its Treasury Digital Fund "currently uses the Ethereum network as the public blockchain for secondary recording of ownership," while "the transfer agent's book-entry records will constitute the official record of the fund and govern the record ownership of fund shares in all circumstances," per the fund's filing effective 25 July 20255. Ethereum as distribution layer and the transfer agent as register is exactly Section 2.1's source-of-truth split. JPMorgan Asset Management put a tokenized money market fund on public Ethereum in December 2025, already covered in Section 1.3.
Ethereum carried $16.69 billion of distributed value across 2,564 assets and 273,597 holders on 20 September 20261. Per Section 4.1's venue list, it also has the deepest set of destinations: Aave Horizon, Morpho, and UniswapX.
Standards are a reason on their own. Ethereum's ERC-20 standard, the token taxonomy paper's Section 3.2 covers, is the baseline every wallet and exchange already supports. Section 2.4 already describes how the ERC-3643 permissioned-token standard checks a holder's eligibility before a transfer; the ERC-4626 and ERC-7540 vaults Section 4.1's Centrifuge profile describes build on the same base layer. An issuer choosing Ethereum inherits tooling, audits, and custodian integrations a newer chain must build from scratch.
Ethereum is also the expensive default. BlackRock's November 2024 BUIDL share classes carry a 50 basis point management fee on Ethereum, Arbitrum, and Optimism against 20 basis points on Aptos, Avalanche, and Polygon, and the release states that "Aptos Foundation, Avalanche (BVI), Inc. and Polygon Labs BD Investments (Cayman) Ltd. each have agreed to pay BlackRock a quarterly fee generally based on the average value of the relevant class of shares for such quarter"6. Issuers still start there anyway, adding cheaper chains later.
The drawback is symmetric with the advantage: every Ethereum transfer is public by default, the privacy problem the permissioned ledgers in the next section were built to remove.
Solana: Throughput, Cost, and Token-Level Compliance
Solana became BUIDL's seventh chain on 25 March 2025, with cross-chain transfers enabled by Wormhole7.
On 20 September 2026 Solana carried 947.1 million of BUIDL's 2,304.7 million units, about 41% of the total and the largest single-network share4.
Solana's own ranking figures are $4.34 billion distributed across 2,799 assets and 634,703 holders on 20 September 20261.
Compliance sits inside the token program rather than in each issuer's own contract. Solana's Token Extensions include a transfer hook, where the "mint requires a CPI to a program implementing the 'transfer hook' interface"; an optional permanent delegate; and confidential transfers that mask balances and transfer amounts while preserving issuer auditability8. An issuer building on Solana gets eligibility checks, forced transfers, and balance privacy as standard-library features rather than custom code.
Backed's xStocks trade on Solana DeFi venues, already named in Section 4.1.
Solana's share of BUIDL is a share of one product, not evidence that institutions in general prefer Solana.
Stellar: Compliance Flags in the Protocol and a Five-Year Head Start
BENJI launched on Stellar in April 2021, Section 4.1's first profile and the fund that put the chain on the map. Franklin's stated reason, from the Stellar Development Foundation's case study and there attributed to Ric Golubov, the Franklin Templeton executive the page names as Vice President, Digital Assets Business Development & Partnerships: "We ultimately chose Stellar because it enabled us to meet these certain token-level controls that met the SEC's strict criteria of operating a mutual fund onchain." The same page states that "Authorization and clawback are native Stellar operations, guaranteed by the chain," that "Transaction fees on Stellar are a fraction of a cent," and that settlement reaches finality "in 5 to 7 seconds, 24/7"9. All three claims belong to the chain's own case study, attributed accordingly, not treated as findings.
The controls sit in the protocol, not a custom contract. Stellar's issuer flags AUTH_REQUIRED, AUTH_REVOCABLE, and AUTH_CLAWBACK_ENABLED let an issuer approve which accounts may hold its asset, freeze that asset, and claw it back, without any application code of its own10.
On 20 September 2026 Stellar carried $3.32 billion of distributed value across only 72 assets and 21,044 holders1.
The chain one fund put on the map now carries several more: Spiko at $1.6 billion, Ondo at $536.2 million, Franklin's Benji at $517.3 million, Realiz at $500.0 million, and Circle at $328.1 million11.
The chain's 2026 signal came from outside crypto entirely. DTCC announced on 27 May 2026 that its tokenization service will connect to Stellar, expected in the first half of 2027, part of "DTCC's standards-driven, multi-chain strategy" following the SEC's December 2025 no-action letter12. What Stellar does not offer is the DeFi venue depth Ethereum has built over a decade.
The EVM Layer 2s, Avalanche, and BNB Chain: Reach at Lower Cost, Often Paid For
Who paid for reach is the clearest story here. BlackRock's November 2024 BUIDL share classes charge Aptos, Avalanche, and Polygon 20 basis points against 50 on Ethereum, Arbitrum, and Optimism, with the three lower-fee ecosystems paying BlackRock a quarterly fee6.
The Arbitrum DAO paid in its own token: its STEP committee recommended 35 million ARB in June 2024, split across six products, Securitize's BUIDL 11 million, Ondo's USDY 6 million, Superstate's USTB 6 million, Mountain's USDM 4 million, OpenEden's TBill 4 million, and Backed's bIB01 4 million, choosing "only 6 products from the shortlist of 17"13.
STEP 2 in May 2025 allocated another 35 million ARB, split 35% to Franklin Templeton's BENJI, 35% to Spiko, and 30% to WisdomTree, and the DAO's own release puts tokenized RWAs on Arbitrum at "over $240 million," a figure attributed to the release14.
What the money bought shows in BUIDL's own supply. On 20 September 2026 Avalanche held 493.4 million units, Aptos 161.5 million, BNB Chain 146.4 million, Optimism 26.4 million, Tempo 17.1 million, Arbitrum 8.6 million, and Polygon 4.5 million4. Put flatly: a subsidized listing is not distribution. Polygon, with the lowest fee tier on offer, held a fraction of one percent of BUIDL's supply, and Arbitrum, which paid 11 million ARB for BUIDL's presence, held under half a percent.
Avalanche's own network figures, by contrast, show $1.68 billion distributed and an $11.40 billion represented figure that rwa.xyz did not attribute to a named platform on 20 September 20261.
A newer reason has nothing to do with foundation money: the distributor already owns the rail. BNB Chain carries $5.74 billion of distributed value and 1.8 million holders1; BUIDL's BNB Chain share class arrived in November 2025 alongside Binance's acceptance of BUIDL as collateral, both covered in Section 4.1.
Base, "the Ethereum Layer 2 blockchain built within Coinbase," carried JPMorgan's JPMD deposit token from 12 November 2025, meeting what JPMorgan calls "growing demand ... for faster and easier money movement on public blockchains," with Coinbase's Jesse Pollak describing "sub-second, sub-cent, 24/7 settlement"15.
Robinhood Chain, "built using the Arbitrum Platform" and operated by Robinhood, reached public mainnet on 1 July 2026 carrying Robinhood's stock tokens for 24/7 trading and use as DeFi collateral16.
On 20 September 2026 rwa.xyz listed a "Robinhood" network with 1,398,259 holders1.
Tempo was announced on 4 September 2025 as "a payments-first blockchain incubated by Stripe and Paradigm"17.
It carried 17.1 million units of BUIDL on 20 September 20264. When the distributor owns the rail, the chain choice belongs to the distributor, not the issuer.
The XRP Ledger is a smaller case. Ondo launched OUSG on it on 11 June 2025 with 24/7 mint and redeem in Ripple's RLUSD stablecoin and "$30M in committed TVL at launch"18.
On 20 September 2026 it carried $456.6 million distributed against $4.05 billion represented1.
Permissioned and Purpose-Built Ledgers: Where the Big Numbers Are
Canton. Digital Asset's July 2024 go-live release describes the network as "an innovative public-permissioned blockchain network designed with the privacy and controls essential to facilitating the exchange of regulated financial assets," naming 32 participants including Broadridge and Tradeweb19.
Under Canton, rwa.xyz files Broadridge's Distributed Ledger Repo at $327.02 billion, the network's only listed platform, against zero distributed value2.
Broadridge itself reported $7.5 trillion processed on DLR in June 2026, an average of $357 billion a day20.
DTCC and Digital Asset announced on 17 December 2025 that a subset of DTC-custodied US Treasuries would be minted on Canton, with DTCC joining the Canton Foundation as co-chair alongside Euroclear21. The label drifted in the meantime: the same December 2025 release's boilerplate calls Canton "the only public, permissionless blockchain purpose-built for institutional finance," eighteen months after Digital Asset's own "public-permissioned." This section reports the function, a validator set of named financial firms with transactions visible only to their parties, not the label.
Kinexys. JPMorgan's April 2026 milestones page calls Kinexys Digital Assets a "private, permissioned blockchain network" and reports it "processing more than $3 trillion in transactions since inception and averaging more than $5 billion daily," a figure covering Kinexys as a whole, payments included, not tokenized assets alone22. JLTXX, the tokenized fund running on it, stood at $810 million on 16 September 2026, already covered in Section 4.1.
The same bank issued JPMD on Base15 and put an asset-management fund on public Ethereum, covered in Section 1.3. The lesson generalizes: chain choice is made per use case, not per firm.
Provenance. Figure's ledger is, in Figure's own filings, "an independent Layer 1 blockchain" that is "public and permissionless," already quoted in Section 4.1, and is built "using decentralized, open-source technology, including the robust framework of the Cosmos SDK and the battle-tested CometBFT," per its developer documentation23.
On 20 September 2026 it held $22.76 billion represented against $83.69 million distributed and 190 holders3. A chain can be public and permissionless in its code and closed in practice; the represented classification Section 2.3 describes is the tell.
Bank-built platforms handle issuance too. The Hong Kong government's roughly HK$6 billion multi-currency digital green bond of February 2024 was issued on HSBC Orion, HSBC's digital assets platform, with the Hong Kong Monetary Authority's Central Moneymarkets Unit as clearing and settlement system, "the first multi-currency digital bond offering in the world" in the government's own words24.
RWA-specific public layer 1s stay small by comparison. Plume carried $185.5 million distributed with 259,674 holders and MANTRA $371,270, both on 20 September 20261, and Ondo's July 2026 decision was that "achieving our objectives doesn't require us to build a chain in the traditional sense at all," already covered in Section 4.1.
Banks land on permissioned ledgers for three reasons in combination: privacy, finality, and validator control, plus the capital treatment the next section explains.
What Regulators Decide for You
Basel. The Basel Committee's cryptoasset standard sorts bank exposures into Group 1, which includes tokenized traditional assets, and Group 2, subject to classification conditions. Classification condition 3 requires that the cryptoasset and "the network on which it operates, including the distributed ledger or similar technology on which it is based, are designed and operated to sufficiently mitigate and manage any material risks," and the standard names "Validating design of the DLT, permissionless or permissioned" among the technology risks a bank must assess25. The standard's own text sets the capital treatment for Group 2 exposures.
The decisive sentence sits in the Committee's 7 December 2023 press release: "The Committee concluded that cryptoassets that use permissionless blockchains create risks that cannot be sufficiently mitigated at present and therefore agreed to retain the existing treatment for such cryptoassets."26
In November 2025 the Committee "agreed to expedite a review of targeted elements of the standard"; as of 20 September 2026 no revised text naming permissionless ledgers had been published27.
SEC. The staff's January 2026 statement is ledger-neutral: "The format in which a security is issued or the methods by which holders are recorded (e.g., onchain vs. offchain) does not affect application of the federal securities laws."28
DTCC's May 2026 announcement that its tokenization service will connect to Stellar, under the December 2025 no-action letter, shows a central depository choosing public chains one at a time rather than by policy12.
EU. ESMA's June 2025 review of the DLT Pilot Regime found that "only three DLT MIs have been authorised" as of 31 May 2025, one of them, 21X AG, running tokenized securities "on a public, permissionless blockchain (Polygon PoS) whereby access to the DLT-TSS and the services provided is permissioned," and recommended making the regime permanent29.
Germany's electronic securities law let Siemens issue a €60 million bond in February 2023 as the "first digital bond on a public blockchain in accordance with Germany's Electronic Securities Act (eWpG)"30.
A €300 million Siemens bond followed in September 2024, "settled via the private permissioned blockchain of SWIAT, and the Trigger Solution provided by the Bundesbank," within minutes and in central bank money31: one issuer, one law, a public chain and then a bank-consortium ledger, both legal.
Regulation narrows the menu for banks and broker-dealers far more than it does for asset managers, which is why the same product family, tokenized Treasuries, appears on Ethereum, on Canton, and on Kinexys at once.
Why Almost Everything Went Multichain
Evidence comes before explanation. BUIDL runs on ten network entries4; Section 4.1 already covers JAAA's eight chains and Centrifuge's "live on 11 blockchains"; ACRED launched on six chains, covered in Section 1.2.
Three reasons explain the pattern. Each chain is a different buyer base and venue set: BNB Chain for Binance collateral, Solana for DeFi venues, the XRP Ledger for RLUSD settlement, all covered in the profiles above. Someone else often paid, as detailed above. And the issuance platform made multichain cheap: on 20 September 2024 Securitize selected Wormhole as "the official blockchain interoperability provider" for "all current and future tokenized assets from Securitize"32, and Section 4.1's hub-and-spoke description of Centrifuge shows the same logic, accounting on one chain, issuance on many.
Multichain carries three costs to match. First, bridge risk: Jump Crypto, the bridge's own backer, wrote in August 2022 that "Wormhole's February exploit took place due to the absence of a validation check that enabled the exploiter to create a false signature and steal over $320M"33; every bridge is a new party in Section 2.4's dependency chain.
Second, throttled bridges and thin tails: Ondo's bridge uses LayerZero's OFT standard to burn and mint USDY across seven chains, verified by three named verifiers, with daily rate limits of 450,000 USDY out and 500,000 in per EVM pathway34.
BUIDL's own Polygon and Arbitrum supply shows what a listing with minimal real distribution looks like4.
Third, the cash leg still has to follow the asset leg: on 14 May 2025 Chainlink, Kinexys, and Ondo executed a cross-chain delivery-versus-payment test settling OUSG on the Ondo Chain testnet against deposit accounts on Kinexys Digital Payments, a test and not production35. Section 4.4 takes up the cash leg directly.
The diagram reads top to bottom as the question an issuer answers before any technical comparison starts. Green marks the public chains, where reach comes from venues and holders already present; red marks the ledgers where reach comes from membership instead. The bottom row is where the compliance logic ends up, the same difference Section 2.4 described as making "compliance part of the token" versus permissioning "the components" themselves. The diagram simplifies in three ways. A firm can sit in two boxes at once, as JPMorgan and Franklin Templeton both do. The same product can appear in three, as BUIDL does on Ethereum, on BNB Chain, and as collateral at Binance. And a "public" label does not settle which row a ledger belongs in, as Provenance and Canton both show.
How to Read a Chain Choice
Five questions, asked in order, read any chain choice in this section or the next one made.
- Who must be able to hold it on day one, and are they already here? Custodians, wallet support, venues, and the investor base the issuer actually wants, per the holder counts above.
- Where does the compliance logic live? In the token contract, ERC-3643 style; in the chain's own token program or asset flags, as Solana and Stellar do; or in the network's admission rules, as Canton and Kinexys do. It is Section 2.4's question, applied to the ledger rather than to the contract.
- Who can see the transfers, and is that allowed? Public by default on every chain in the public-chain profiles above; privacy and controls are the product the permissioned ledgers sell instead.
- What does the ledger cost, including capital? Fees and fee tiers, integration and bridge exposure, and, for a bank, the prudential treatment of the ledger itself.
- Who is paying for this deployment, and what happens when they stop? Foundation and DAO subsidies end; watch supply on the subsidized chain, not the announcement.
A chain that answers all five well for one product may answer them badly for the next, why the same issuer often runs on three.
Bridge to Section 4.3
Every chain choice documented above was made by a named firm and then executed by a named custodian, transfer agent, and, in the multichain cases, a bridge operator. The chain never chose itself, and none of the firms above made the choice once and walked away from it: BlackRock and Franklin Templeton both revisited the decision as their products grew and as new chains offered new terms. Section 4.3 covers the players who carry out that choice day to day: issuers, custodians, SPVs, and investors, and what each one is responsible for when the token and the off-chain record disagree.
- Chain choice is a distribution decision made after the legal-right decision, and it is read from where holders, custodians, and venues already are: Ethereum's 273,597 holders on $16.69 billion and BNB Chain's 1.8 million holders on $5.74 billion are different distribution strategies, not accidents.
- The distributed market is Ethereum-plus-EVM with three non-EVM anchors, Solana, Stellar, and the XRP Ledger, while about 95% of represented value sits on two purpose-built ledgers, Canton and Provenance.
- Public chains differ mainly in where compliance lives, in the contract, in the chain's own token program, or in issuer-set flags, and in cost; several deployments were paid for by chain foundations or DAOs, which bought listings rather than holders, as BUIDL's Polygon and Arbitrum supply shows.
- Banks choose permissioned ledgers for privacy, validator control, and the Basel treatment of permissionless chains, and the same bank uses public chains where the use case allows it, as JPMorgan does with Kinexys alongside Base and public Ethereum.
- Multichain is the compromise between reach and control, and every bridge is a new party in the dependency chain; its costs are bridge risk, throttled transfer limits, and a cash leg that must still follow the asset leg.