What tokenized private credit actually funds

Executive summary

Two numbers describe this market and they are not additive. On 25 July 2026 RWA.xyz's credit page recorded $6.99bn of distributed value across 2,512 assets and 186,390 holders, alongside $35.94bn of represented value, meaning a conventional asset with a record written to a chain. The gap between the two is where most of the reported growth in tokenized credit has come from, and it is not lending that has been changed by tokenization in any economic sense.

The largest entry in the category is a represented one. Figure's HELOC token stood at $17.05bn on the same reading, with an inception date of 17 January 2022, no published yield, a blank underlying-assets field and zero transfers in the preceding month. The second, the DOM X Arizona copper-gold project, was announced on 23 April 2026 at $11bn or more. Their combined value exceeds the whole distributed figure four times over, so neither sits inside it.

What is left, once the represented stock is set aside, funds four things: prime US home equity, secured financing against liquid collateral for crypto trading firms, tranches of broadly syndicated leveraged loans, and bitcoin mining capital expenditure. The Financial Stability Board's report of 6 May 2026 defines private credit narrowly as "nonbank direct lending to medium-sized companies negotiated on a bilateral basis" and sizes it at $1.5trn to $2trn at end-2024, with roughly $1trn in the United States. Almost none of the tokenized stock matches that definition, and the $6.99bn is about 0.4% of the FSB's midpoint in any case.

Key findings

1. The largest asset in the category has never moved. The Figure HELOC token was valued at $17.05bn on 25 July 2026, classed as a represented asset on Provenance, with $0 of monthly transfer volume, zero transfers recorded in the month, no management or performance fee and no disclosed underlying assets. It is a loan tape with a hash, not a transferable claim.

1. Where loan-level disclosure is genuinely good, the borrower is a prime US consumer. Figure's Amendment No. 5 to Form S-1, filed 10 September 2025, reports over $16bn of home-equity lines originated from late 2018 to 30 June 2025 and roughly $6bn in the twelve months to that date, with an average FICO score of 755 and average borrower income of approximately $186,000 in the six months to 30 June 2025, combined loan-to-value capped at 90% and debt-to-income at 50%. That is well-documented consumer secured credit. It is not corporate direct lending.

1. The largest transferable pool is secured crypto lending priced at 4.87%. Syrup USDC held $1.04bn on 25 July 2026, down 19.42% over thirty days, at a trailing seven-day yield of 4.87% against a thirty-day figure of 5.30%, with a net asset value of $1.17, zero management, performance, subscription and redemption fees, daily first-in-first-out redemption and a US domicile carrying SEC central index key 0001995469. Maple describes the underlying loans as overcollateralised with automatic margin calls and states "zero losses to date"; it publishes no collateral composition.

1. The best-documented credit in the category is not private credit. The Janus Henderson AAA CLO Fund held $689.7m on 25 July 2026 across eight networks, with 33 holders, a 0.50% management fee, a British Virgin Islands domicile, Anemoy as issuer with Janus Henderson as sub-adviser, custody at J.P. Morgan, and trailing yields of 2.99% over seven days and 3.71% over thirty. Its exposure is senior tranches of syndicated loans, and Centrifuge announced on 25 June 2025 that Sky had committed $1bn to it through Grove.

1. The largest headline number in the category rests on no published valuation. The Bridgetower release of 23 April 2026 describes tokenizing $11bn or more of securities from the DOM X Arizona copper-gold project using Chainlink's cross-chain protocol, proof-of-reserve, valuation feed and runtime environment. It names no auditor, no reserve or resource report, no security type, no domicile and no supervising regulator.

Analysis

Start with what the label should mean. The FSB uses a narrow definition and explains why: there is "no common definition" and that alone frustrates measurement. Its borrower profile is specific. Private credit borrowers carry 5 to 6 times debt to EBITDA, higher than the roughly 4 times observed in leveraged loans and closer to 7 times once accounting adjustments are stripped out; payment in kind appears in about 12% of loans, with toggles accounting for about half, and its use has risen since 2022 without sign of decline; spreads run from around 230 basis points on business development company loans to around 350 basis points on private credit CLO investments and 200 to 400 basis points on net-asset-value facilities.

Now read the tokenized pools against that. A pool paying 4.87% over the last seven days, with a first-in-first-out redemption queue that clears inside a day, is not being paid to hold the default risk of a 6-times-levered mid-market borrower on a covenanted bilateral loan. It is being paid to lend against liquid collateral, at a margin call trigger, to counterparties whose business is leverage. That is a coherent and probably useful business. It is also, in credit terms, closer to repo than to direct lending, and the risk it carries is the risk that collateral gaps through the liquidation threshold in a fast market rather than the risk that an operating company misses an interest payment. A CLO fund yielding 2.99% is a third thing again: rated, tranched, senior exposure to a public loan market, wrapped offshore and distributed onchain.

Grouping the three under one heading obscures the only question that matters to an allocator, which is what has to go wrong before the money is not returned. It also obscures a circularity. Sky's $1bn commitment through Grove means reserve assets standing behind a stablecoin were routed into senior CLO tranches. That is a defensible reserve decision, and it means the demand for tokenized corporate credit is in part demand from other tokenized liabilities rather than from end investors reallocating out of conventional funds.

Documentation quality falls into four tiers, and the drop between them is steep. At the top sits the issuer that files with the SEC: Figure's prospectus carries origination volume, borrower credit quality, underwriting caps and the statement that all loans and related data other than borrower personal information are recorded on a public chain. Below it sits the offshore fund wrapper with a named institutional manager and custodian, where the structure is disclosed and the holdings are not; the Janus Henderson fund's underlying-assets field on the tracker is empty, and 33 holders for $689.7m means the register is a handful of balance sheets. Below that sits the protocol pool, where domicile, fee schedule and redemption mechanics are public but the custodian, administrator, auditor and eligible-investor list are behind a login and the collateral schedule is not published at all. At the bottom sits announcement-grade tokenization, where an $11bn figure enters a public tracker on the strength of a press release.

Poor disclosure has already produced losses in this market, and the failure was not analytical. Goldfinch's $20m USDC loan to Stratos, a four-year facility at 11% originated in February 2022, was written down by $7m by October 2023. Of the money, $5m went to a real-estate technology firm that stopped paying, $2m went into digital-asset positions that the underwriter said it "was not aware" of at the time of investment, and $13m went to an e-commerce company that performed. What broke was use-of-proceeds control and reporting: the lender did not know what its loan was funding until after the fact.

The current stack has answered that mostly by retreating from unsecured lending into overcollateralised lending and fund wrappers. That is a real improvement in loss experience and it is also an admission. The specific promise of tokenized private credit was that a public ledger would let capital reach borrowers whose credit conventional lenders could not price. The market that exists instead lends against collateral it can seize, buys tranches somebody else has rated, or records loans it was already making. The three places disclosure is worst are precisely the places where that promise would have to be tested: the collateral schedules behind the secured pools, the holdings behind the offshore credit wrappers, and any valuation basis at all behind resource-project tokens.

Methodology and data

Category and asset figures are RWA.xyz readings taken on 25 July 2026. The provider separates distributed value, the portion of an asset issued and transferable as a token on an indexed network, from represented value, a conventional holding with a chain record; the two are reported side by side and are not additive. Asset-level values, yields, fees, domiciles, holder counts and inception dates are taken from individual asset pages rather than from the category listing where both were available.

The composition estimate sums the eight largest identifiable distributed entries on the category page, using asset-page values for Syrup USDC and the Janus Henderson fund and page listings for Blockstream Mining Note 2 at $750m, PKH Mining Note 2 at $577m, VuMe Bond 2030 at $500m, PRIME at $452m, the Securitize AAA CLO tokenized fund at $354m and Syrup USDT at $345m. Those eight total about $4.71bn, or roughly 67% of the $6.99bn. The same page ranked STOKR first among platforms with four assets at $1.3bn and Centrifuge third with four assets at $748m.

Yields quoted as seven-day and thirty-day figures are the tracker's trailing calculations, not contractual coupons, and they move with both the underlying portfolio and the token supply. Market-size comparisons use the FSB's end-2024 range. Loan-level figures for Figure come from the S-1 and refer to the periods stated there, not to 2026.

Limitations

The data this piece most needed is loan-level and it is not public. No tokenized credit product examined here publishes a collateral schedule, an obligor list or a maturity ladder outside a login. The FSB says the same thing about the conventional market in its own terms: the key challenges are "the lack of harmonised private credit definitions, and limited granular fund-and loan-level data", and its members captured around $220bn of bank credit lines against commercial estimates more than twice as large.

The aggregates do not reconcile with themselves. The category page and the Figure asset page did not agree on that asset's value during this research, and the asset-page figure of $17.05bn is used here. The same page attributed two assets worth $967m to Maple on a day when the tracker's own page for Syrup USDC alone showed $1.04bn. Read at this resolution the totals carry an error bar wider than the differences between the products.

Valuation timing is a specific hazard for tokens. The FSB notes that private credit valuations are "updated infrequently, often on a quarterly basis, which may be adequate in normal market conditions but less so under stress", that managers hold significant discretion, and that stale marks create a first-mover incentive. A token that transfers continuously against a net asset value struck on a slower cycle inherits that incentive and sharpens it.

Loss history proves less than it appears to. Maple's statement of zero losses to date is the issuer's own, unaudited, and covers a period the FSB describes as untested: private credit as a whole "has not been tested during a severe economic downturn, which could expose leverage and borrower credit quality vulnerabilities". A collateralised book with automatic margin calls has a specific failure mode, a gap through the liquidation level, and nothing in the public record measures how much of these pools could be liquidated at once.

Two figures could not be obtained. Figure publishes no weighted average coupon for its home-equity pools in the prospectus, so this piece cannot state at what rate those loans were made; and the securitisation rating reports that would carry it were not accessible. Nor is there any public basis for the DOM X valuation, which means an $11bn number now sits inside a market aggregate that others will cite.