Written by: Vaidik Mandloi
Compiled by: Block unicorn
In the 1970s, Bruce Bent and Henry Brown created the first-ever money market fund. The idea was remarkably simple. Due to a regulation from the Great Depression era, the interest rate on savings at U.S. banks was capped at 4.5%. Although U.S. Treasury yields were over 9% at the time, the minimum investment to purchase Treasuries was as high as $10,000. Bent and Brown decided to pool small deposits, buy Treasuries on a large scale, and return the profits to investors. Today, money market funds have reached a scale of approximately $8 trillion.
Stablecoins are engaging in a similar practice, only this time the assets are private credit, a market worth $2 trillion where a minimum of $1 million is required to enter. Yield-bearing stablecoins are being used to pool small deposits and channel them into the credit space.
Today, I will delve into how this is happening and how Goldfinch (the first-ever attempt to conduct such operations with real funds) collapsed, resulting in $56 million of depositor funds being trapped in Kenyan motorcycle loans.
How Stablecoins Became Money Market Funds for Private Credit
In the 1990s, U.S. banks provided about half of the debt capital to businesses and consumers, but today that figure is around 20%. This is because after 2008, new capital rules took effect, making it too costly for banks to hold leveraged loans. Consequently, banks entirely withdrew from the mid-market lending business, and private credit funds took their place.
Apollo, Blackstone, and KKR raised funds from pension funds and insurance companies and began lending to companies abandoned by banks, charging high premiums because these borrowers had no alternative.
The market has grown from less than $200 billion in 2008 to over $2 trillion today, and almost all of this capital comes from institutional investors writing checks of $5 million or more.
One of the main reasons for the million-dollar minimum threshold in private credit loans is their high management complexity. Each transaction requires due diligence, restructuring, and monitoring over several years. Managing a fund with ten institutional limited partners (LPs) each contributing $50 million is far easier than managing a fund with retail investors each contributing $500, and even then, scaling investments can often be unprofitable. This is why, over the past decade, only pension funds and insurance companies could access such yields, typically ranging from 8% to 12%.
It is at this point that yield-bearing stablecoins have changed the game, just as Bent and Brown opened access to Treasuries in the 1970s. Although the paperwork is still handled by institutions, with funds like Apollo underwriting and managing the risks, now tokenized sub-funds can accept deposits of any size and channel them into institutional strategies without having to manage thousands of individual investors.
Apollo recently launched the tokenized fund ACRED, whose diversified credit fund has attracted $109 million in inflows. Investors can even use it as collateral on the Morpho platform for margin lending and re-investment, thereby achieving leveraged yields.
Figure has built an entire on-chain lending system, has issued $21 billion in loans, successfully listed on Nasdaq, and launched YLDS—a yield-bearing stablecoin with a circulating supply of $376 million. Other protocols, such as Pyse and Glow, go even further by tokenizing solar projects, allowing investors to fund solar installations in developing countries with a few hundred dollars and earn an annual percentage yield (APY) from monthly electricity fees.
This does not mean that the minimum investment requirement for the fund itself disappears. The ACRED fund still requires a direct investment of $5 million. However, once the fund is tokenized, its tokens can be traded on secondary markets with no minimum investment limit, and they can interoperate with DeFi (Decentralized Finance) systems in ways that traditional fund shares cannot match.
In traditional private credit, your funds are locked for years, with quarterly redemptions capped at 5%. On-chain, however, funds can be flexibly combined and are liquid 24/7. For companies like Apollo and Figure, this allows them to access a $315 billion pool of stablecoin capital actively pursuing yield. By tokenizing funds, they can directly tap into this capital pool, opening new distribution channels without having to build retail infrastructure from scratch.
A year ago, the total on-chain private credit was just $400 million; today, it stands at $5.87 billion, a 15-fold increase in 12 months. However, this still represents only 0.30% of the $2 trillion global private credit market. Half of all new stablecoin supply in the first quarter of 2026 came from yield-bearing stablecoins, meaning most new stablecoin capital is now chasing active yield, not just a dollar peg.
Moreover, because every dollar of on-chain credit can be used as collateral and recycled through DeFi protocols, the actual financial activity it generates is a multiple of the dollar amount.
Take ACRED as an example. An investor deposits $10,000 on Morpho, borrows $7,000 in USDC using this deposit as collateral, then uses this USDC to purchase more ACRED and deposits it again as collateral. This way, a single deposit can generate over $17,000 in credit exposure. In contrast, traditional private credit: the same $10,000 would sit idle in a fund for 5 years, earning no yield. On-chain, this compounding effect occurs simultaneously across multiple layers, which is why the ACRED market is growing faster than its raw dollar size suggests. However, it also means that if the underlying loan defaults, losses propagate through every layer of the cycle.
Tokenization does not mean the underlying risks are reduced. Often, these risks are overlooked as capital continuously flows in, providing enough funds to cover redemptions. But as inflows slow, the gap between token promises and actual loan values begins to show. Investors try to exit, but liquidity is insufficient, or the token price becomes disconnected from its intrinsic value.
Something similar happened with Goldfinch; it was one of the first protocols to put private credit on-chain when launched in 2021, but recently had to shut down with $56 million of depositor funds stuck in Kenya and Nigeria.
What Went Wrong with Goldfinch?
Goldfinch raised $25 million from a16z in 2021 to channel cryptocurrency funds, which at the time were only yielding 2% to 3% in DeFi lending pools, to businesses across Africa and Southeast Asia. Borrowers in these regions had to pay interest rates of 15% to 25% because local banks were unwilling to serve them.
The concept was to allow anyone holding USDC to deposit into Goldfinch pools, and then smart contracts would allocate funds to respective borrower accounts within seconds. However, underwriting a loan for a motorcycle financing company in Nairobi meant someone had to understand the Kenyan transportation economy, personally verify the borrower's accounts, and potentially visit the borrower's office in person if repayments stopped.
But on the blockchain, none of these things are possible. Once USDC is converted into Kenyan shillings and invested in a loan book, depositors have no way of knowing how their funds are being used, the financial status of the borrower, or even whether the loan terms are being honored. All critical information related to loan performance has left the blockchain and is in the hands of borrowers in countries most depositors have never visited.
This is why it took months for anyone to notice that Tugende Kenya, without authorization, transferred $1.9 million of its $5 million loan facility to Tugende Uganda in 2022. Nearly 40% of the loan was diverted to another legal entity in another country. Meanwhile, depositors continued to receive what they believed was 10% to 12% interest, completely unaware that the principal backing their yields had gone to a place never mentioned in the loan agreement.
If a traditional private credit firm discovered such a severe breach, it would recall the loan and force a restructuring within days. But Goldfinch depositors learned about this through posts on the company's governance forum, with their only option being to vote on a proposal that held neither legal authority to seize assets nor to audit the remaining assets.
By 2023, Tugende had completely defaulted and vanished. Over its $113.3 million lifespan, Goldfinch issued 24 asset pools, of which only 13 were fully repaid. The remaining 8 pools have $53.82 million in outstanding loans, and none are performing under the original terms. Most pools are undergoing restructuring, with repayments of less than $51,000 per month per pool, meaning recovering the full $53.82 million would take 8 to 15 years at this rate.
Goldfinch took on all the risks of emerging market currency volatility and limited credit history, yet possessed almost none of the infrastructure that traditional lenders spent decades building and managing to mitigate these risks. For example, banks offering loans in Kenya have local offices and regulatory relationships, giving them greater leverage when deals go wrong.
But Goldfinch funneled funds from anonymous global wallets to the same type of borrowers without any such supporting structures. This made the information gap between lenders and borrowers even larger than in traditional transactions and left depositors with almost no ability to intervene when things fell apart.
On-chain fund transfers constitute only about 10% of the processes required for lending. The remaining 90% is underwriting and fund recovery, which is highly localized and costly. Such underwriters need to establish a baseline of credibility for the entire asset class, which itself is still fighting for its right to exist. Every dollar lost in underwriting makes it harder for the next institutional partner to go on-chain and degrades the credibility of the entire asset class. The difficulty in credit has nothing to do with on-chain operations, and anyone building in this space without understanding this is essentially building the next 'Goldfinch.'
