Where tokenised private credit has lost investors money

What is tokenised private credit?

Tokenised private credit is an investor’s claim on private loans, recorded as a token on a blockchain instead of as a line in a fund administrator’s database. It is one use of tokenisation, the practice of putting fund interests and other assets on a blockchain. The loans stay where they were made, in a fund or a lending pool, and the borrowers owe the same money on the same terms.

Two designs sit under the name, the feeder fund and the on-chain lending pool. Securitize announced on 30 January 2025 a feeder fund called ACRED, which gives tokenised access to the Apollo Diversified Credit Fund. Securitize describes that underlying fund as a “non-traded, closed-end interval fund” whose strategy spans corporate direct lending, asset-backed lending, and performing, dislocated and structured credit. An investor buys the feeder, and the feeder buys the underlying fund.

In an on-chain lending pool, lenders deposit crypto assets, usually stablecoins, and a protocol lends them to borrowers. The Block reported in December 2022 that Maple Finance’s pools were run by delegates, which in its words “processed due diligence for investors”. Orthogonal Trading’s credit arm acted as one, and a separate company, M11 Credit, ran the M11 USDC pool. Maple’s homepage now offers “institutional-quality yield from overcollateralized loans”, unlike the undercollateralised loans The Block described in 2022, open to non-US investors, and says loans, collateral and transactions are verifiable on the blockchain.

Neither design puts the loan itself in the investor’s hands. The investor holds a claim on a fund or a pool, and the token records that claim.

Who holds the loans and who keeps the register?

In a tokenised feeder fund such as ACRED the underlying fund holds the loans and a named transfer agent keeps the register that the token represents; in an on-chain pool the pool holds the loans and the blockchain carries the record. For ACRED, Securitize’s product page (read on 1 November 2026; its assets under management are dated 21 October 2026) names Securitize Capital, LLC as general partner, Securitize Markets, LLC as placement agent, Securitize, LLC as transfer agent and Securitize Fund Services, LLC as fund administrator, and says the fund “invests in the Apollo Diversified Credit Fund”. The investor can choose between self-custody and holding the investment in a Securitize account.

Question	ACRED, a feeder fund	On-chain lending pool
Who holds the loans?	The underlying Apollo Diversified Credit Fund, which the feeder invests in (Securitize, 2025 and 2026)	The pool, which lends to borrowers; at Maple a delegate ran due diligence (The Block, 2022)
Who keeps the register?	Securitize, LLC, as transfer agent (Securitize, 2026)	Recorded on the blockchain; Maple says loans and transactions are verifiable on it (Maple, undated)
What does the investor hold?	Interests in a feeder limited partnership, held as a token (the page calls them both shares and LP units)	A claim on the pool, as a token
Who may buy?	Accredited investors, from $50,000 (Securitize, 2026)	Set by each protocol; a Goldfinch proposal author wrote that Goldfinch Prime “was onchain and for non US citizens” (Goldfinch governance forum, June 2026)

The token gives the investor a register entry that can sit in the investor’s own wallet, and that can be transferred where the issuer allows it. RWA.xyz, a data provider, separates the two cases in a framework it published on 21 November 2025. Distributed assets “can be moved to wallets outside the issuing platform”, though transfers may be limited by whitelist or eligibility controls. Represented assets cannot be moved outside the issuing platform, “whether due to design or regulatory constraints”.

ACRED’s token can move to a self-custody wallet, and Securitize’s page states that “No secondary market exists for the Fund’s shares though the Advisor will seek to develop one.” A holder can carry the record but has no listed buyer. An LP used to a capital account statement from a fund administrator is getting a similar record from a different keeper.

How big is the tokenised private credit market?

On RWA.xyz’s tokenised credit page, read on 1 November 2026, distributed value was $8.33 billion and represented value was $41.62 billion, across 2,724 assets and 205,585 holders. The page describes its scope as all non-sovereign debt, such as private credit, onchain lending and corporate credit, so the total is wider than private credit funds. It shows the date as “11/01/2026”, in the month-first format of its own transaction dates, which run from 10/25/26 to 10/30/26.

The page does not define its two headings. RWA.xyz’s framework post does, and under its definitions the larger figure is the one the holder cannot take out of the issuer’s platform. The post names Figure’s tokenised home equity lines of credit as an example of represented assets, which operate within Figure’s own lending marketplace. Most of the value counted as tokenised credit is therefore recorded on a blockchain but held inside its issuer’s own platform, and the $8.33 billion is the part that could in principle move.

The league table on the same page lists STOKR first at $1.5 billion and Maple Finance second at $969.8 million, with Securitize sixth at $423.1 million. The table has a Distributed, Represented and All toggle, and the page does not label which view it shows by default.

Sources also disagree on a single platform. Maple’s own site lists assets under management of $4.43 billion, without a date, which is a vendor claim and more than four times the $969.8 million that RWA.xyz shows for Maple. The two numbers may count different things, and neither page says which. Maple describes its business as “digital asset lending and yield products” and “permissioned secured lending for sophisticated allocators”.

For a single fund the scale is smaller still. Securitize’s ACRED page showed assets under management of $34.65 million, with the note “Data as of 10/21/26”.

Where have tokenised private credit investors lost money?

Two documented cases are Maple Finance in 2022 and Goldfinch from 2023, both on-chain lending pools where borrowers defaulted.

In December 2022, The Block reported that Orthogonal Trading had defaulted on eight loans totalling $36 million on Maple Finance, which it said affected “some 30% of active loans” on the protocol. M11 Credit said Orthogonal Trading had said several times during November that it had only $2.5 million of exposure to the FTX exchange. On 3 December, The Block wrote, the firm said it was defaulting because much more of its money was stuck there. Orthogonal Trading did not respond to The Block. Remaining investors in the M11 USDC pool faced a hit of roughly 80%, and investors in the M11 WETH pool faced 17%. Maple expected to recover at least $2.5 million. The Block described Maple as a protocol that offers institutional investors undercollateralised loans, and Maple’s Sid Powell said the platform may look to introduce partially collateralised loans.

Goldfinch is the second case. TechCrunch reported in September 2023 that Tugende, a motorcycle-taxi financier, took a $5 million loan from Goldfinch in October 2021, paid $53,400 of interest a month until May 2023 and defaulted from June 2023. Goldfinch found during Tugende’s quarterly reporting in December 2022 that $1.9 million meant solely for the Kenyan entity had gone to its Uganda operations, without the Goldfinch community’s consent. Warbler Labs, the company building Goldfinch, told investors in February. Warbler projected a write-down of 3.95% on the net asset value of Goldfinch’s Senior Pool, and later said a restructuring could bring that below 0.79%.

The later record is thinner. CoinPaprika, on 30 August 2026, reported that a Goldfinch vote on 23 June 2026 approved a wind-down of Goldfinch Prime and put the protocol into maintenance mode, and Phemex News reported the same vote on 24 June. The proposal itself, GIP-87, posted on Goldfinch’s governance forum on 12 June 2026, says that “a number of borrower pools experienced serious performance issues” and that the original protocol “enabled approximately $100M of loans”. In that thread on 7 July 2026, a proposal author wrote that Prime’s tokens were automatically redeemed at full value, with one extra month of interest. CoinPaprika adds that $56.15 million is still owed on the legacy pools, a figure it takes from DefiLlama. DefiLlama’s own data shows that amount borrowed on 23 June 2026, the day of the vote, and zero from 25 June, without saying whether that reflects repayment or a change in what it tracks.

Final losses are not known. A proposal author wrote in June 2026 that recovering the legacy pools’ payments “could take another 2 years or more”, and the thread gives no loss percentage. One forum commenter, not one of the proposal’s authors, wrote that the protocol had “lost people thousands of dollars of savings”, the commenter’s own included.

These are two documented cases, and no source gives a loss rate for the category. Both come from crypto-native lenders. No documented default in a tokenised feeder of a conventional credit fund such as ACRED turned up in the research for this page, which says only what was found.

What is the downside of tokenisation for private credit?

The downside of tokenisation for private credit is that the token can move faster than the cash behind it, and that the issuer’s page does not say what the holder can claim if the platform fails. This is Unmatched Capital’s analysis of the sources above and not a finding of any of them.

Securitize’s launch release for ACRED said it offers “native redemptions at a daily net asset value”. Its product page says redemptions are quarterly, that at least five percent of the underlying fund’s shares are offered for repurchase each quarter, and that full redemptions may be delayed.

Full redemptions are not guaranteed and may be subject to delays due to market conditions or blockchain settlement mechanics.

A daily price and a quarterly exit may describe different things, and neither page reconciles them. A holder faces the underlying fund’s repurchase queue whatever speed the token travels at. In a real estate syndication the exit is set by when the property sells; here it is set by the fund’s repurchase offers.

On cost, the ACRED page gives an estimated expense ratio of 0.72%, which its footnote says includes the 0.5% management fee and operating expenses. It lists separately the underlying fund’s Class I expenses of 3.34%, as of 31 December 2024, and a withholding tax impact of 0.50%, which the 0.72% excludes. It does not say whether the underlying fund’s expenses come on top of the 0.72%, so the all-in cost of holding the token cannot be read from it.

On reporting, Goldfinch’s loans were visible on a blockchain, yet its own wind-down proposal in June 2026 said only that “a number of borrower pools experienced serious performance issues”, and the thread gives no loss figure. A visible loan book is not the same as a valued one.

The legal claim is the least documented part. The ACRED page names Securitize affiliates as the fund’s general partner, placement agent, transfer agent and fund administrator, but has no text on what a holder can claim if Securitize, the transfer agent or the platform fails, or on how ownership of the token relates to the LP units. Staff of the US Securities and Exchange Commission set out two cases in a statement of 28 January 2026. Where an issuer tokenises its own security, the issuer or its agent “maintains the master securityholder file on one or more crypto networks”. Where a third party tokenises a security issued by someone else, the token “may or may not” confer rights as a holder of that security, and holders may be exposed “to risks with respect to the third party, such as bankruptcy”. The ACRED page does not say which case fits its token, and the offering documents are where to look.

The case for tokenisation is real. A token in the investor’s own wallet, a transfer agent named on the record and loan data that Maple says anyone can verify are genuine changes, and Goldfinch Prime’s tokens were redeemed at full value when it closed, though not the legacy pools. None changes the borrowers’ credit risk or the fund’s repurchase terms.

Who can buy tokenised private credit, and under which rules?

ACRED is open to accredited investors only, with a $50,000 minimum, according to Securitize’s page, and RWA.xyz’s page for the token lists US accredited investors and non-US professional investors as eligible; and its launch release says qualifying investors can buy exclusively through Securitize Markets. Goldfinch Prime and Maple’s overcollateralised yield product were aimed outside the United States: a Goldfinch proposal author wrote that Prime “was onchain and for non US citizens”, and Maple’s homepage offers its product to non-US investors.

In the European Union, ESMA’s page on the DLT Pilot Regime says Regulation (EU) 2022/858 has applied since 23 March 2023 to trading and settlement systems for tokenised instruments. It lists shares below EUR 500 million of market capitalisation, bonds and other forms of securitised debt below EUR 1 billion of issuance, and UCITS funds with less than EUR 500 million under management, and says the list may be extended. A tokenised credit note would fall under the securitised-debt line. How a tokenised private credit fund is treated is left open there, and a reader should not infer it.

In the United Kingdom, the FCA published policy statement PS26/7 on 30 April 2026. It finalises Handbook guidance for fund managers who want to keep the unitholder register on distributed ledger technology, and new rules on dealing directly with the fund, for UCITS management companies, managers of authorised alternative investment funds and depositaries of authorised funds. Private credit is not mentioned, and neither is who may invest in a tokenised fund. Its scope lists only managers and depositaries of authorised funds, and it says the paper may also interest those of unauthorised funds.

In Singapore, the Monetary Authority of Singapore’s managing director said in a keynote on 9 October 2026 that the authority’s Guide on the Tokenisation of Capital Markets Products “explains how existing securities laws apply”, and that the same activity and risks should meet the same regulatory outcomes whatever the technology. Who may buy a tokenised credit fund is outside what the speech covers.

In Switzerland, the DLT Act has been in force since 1 August 2021. FINMA’s release of 18 March 2025, on licensing the first DLT trading facility, describes DLT securities as “book-entry securities” that are “transferred and held on a blockchain”. The release says the facility’s offer is aimed only at supervised participants, usually banks, and investors in tokenised credit funds do not feature in it.

Across the five regimes, the pages read treat the token as a way to record or trade an instrument that existing law already governs, and none says the token widens who may buy. Eligibility still follows the fund’s own documents and the securities rules where the investor is offered the fund: ACRED is open to US accredited and non-US professional investors, while Goldfinch Prime was for non-US citizens and Maple offers its yield product to non-US investors. Hong Kong, Dubai and other centres were not read, so that conclusion rests on five.

What to check before buying tokenised private credit

Before buying tokenised private credit, check what the token claims, who keeps the register, and how and when cash comes back. Securitize’s ACRED page shows why: its token is the record, while cash returns through quarterly repurchase offers of at least 5% of the underlying fund’s shares. Seven questions cover it:

1. What does the token represent: a share of a fund, a claim on a pool or a note, and who is named as transfer agent?
2. What can the holder claim if the issuer, transfer agent, platform, smart contract or the holder’s own wallet fails?
3. Can the token leave the issuer’s platform, in RWA.xyz’s terms is it distributed or represented, and is there a secondary market?
4. What are the redemption dates and limits, and what happens if requests exceed them?
5. What collateral do borrowers post, and who carried out the due diligence on them?
6. Who reports losses, how often, and is the loan book valued or only visible?
7. Which law governs the fund, which rules apply to an offer made where you live, and do the fund’s documents say you are eligible?

The editorial position of this page is that tokenised private credit should be judged first as a lending product, on its exit terms and loan reporting, then on the token’s legal and settlement terms. In both documented cases the losses came from lending. None of this is investment advice.