Tether and Fasanara Capital launched a private credit fund called StableFund on Wednesday, seeded with $400 million of the two firms’ own money and targeting as much as $3 billion from institutional investors. It is an evergreen fund using USDT as settlement infrastructure for short duration, asset backed lending to businesses and consumers through fintech platforms in more than 60 countries. Fasanara, a London manager with over $6 billion under management, runs the investments; Tether sources the financing and moves the money on and off chain.
Nothing broke here. Nobody got rugged. An established credit manager and the most profitable company in crypto are putting real capital into trade receivables and supply chain finance, boring productive lending that banks have spent two decades retreating from. Good.
It is also the fifty-sixth time this series has found the same seam. One half of this arrangement you can verify to the byte. The other half you cannot check at all. And the first half is about to be used to sell the second.
TL;DR
- StableFund launched 9 September 2026: $400M seeded, $3B target, USDT as the settlement rail for asset backed lending in 60+ countries.
- The settlement leg is verifiable by anyone with a block explorer. The loans behind it are verifiable by nobody outside the fund.
- Evergreen means no maturity date, so cash never settles the argument about the marks. Everyone subscribes and redeems at a NAV that is a model output.
- Fast settlement does not compress credit risk. It widens the gap between the verifiable event at the front and the unverifiable outcome at the back.
- On-chain gaming can be fully verified because it has nothing off chain to be wrong about.
What is actually on chain
The payment leg, and only the payment leg. When USDT moves from the fund to a fintech lender in Manila you get an amount, two addresses, a block and finality, and a hostile stranger can check every one of those facts in four years without asking permission. That is a proof: falsifiable by you, mechanically, and loud when it fails.
What is not on chain is the loan. A ninety day invoice owed by a wholesaler in Lagos, a consumer instalment in São Paulo, repaid or not by an obligor who has never heard of a block explorer. USDT settlement does not move the probability of repayment by a basis point. It moves how fast the money arrived.
The number doing all the work
Evergreen is the word to sit with. A closed end fund lends, harvests and liquidates, and the final distribution settles every argument about the interim marks. The cash arrives or it does not. Truth is scheduled.
An evergreen fund has no such moment, by design. Investors subscribe and redeem continuously at net asset value, and NAV on a book of private loans is not a price, because there is no exchange, no order book and no bid. It is a mark: a model output built from a discount rate, an expected loss curve, an arrears assumption and a house view on a receivable forty days late in a country where that is normal. Every input is defensible. Every one is also a choice, made by or for the party whose fee is calculated from the answer.
That is not an accusation. Every private credit fund on earth works this way and the careful ones do it conservatively. The point is category, not conduct. One side of this fund is a number a stranger can verify in a browser tab. The other nobody outside the building can falsify at all. Both go into the same report, in the same font, and you are invited to feel the same way about each.
The test this series keeps running
Ask what it would take for you, with no special access, to show the number is wrong. For the USDT transfer: a transaction hash and thirty seconds.
For the NAV: the loan tape, the obligor identities, the arrears buckets and the recovery assumptions. All of it sits with the manager, none of it is owed to you, and having it would only start an argument about judgement. There is no answer, because the method is the product. Same shape as a closed casino telling you its RNG has been certified. “Our randomness is audited” and “this fund is worth $412 million” are the same category of object: a conclusion produced by a process you are not allowed to rerun.
Speed widens the gap
The rail is a real improvement, and moving working capital to a frontier market lender without three correspondent banks and a two day float is the strongest argument stablecoins have. But credit risk runs on the borrower’s calendar. Settlement went from days to seconds; learning whether a ninety day receivable pays still takes ninety days. The verifiable event now happens instantly at the front while the unverifiable outcome resolves quietly a quarter later, and that ordering is how credibility passes from the half that earned it to the half that did not.
What this has to do with a coinflip
The structural version, not the promotional one. In an on-chain raffle built properly, the thing being settled and the thing being verified are the same object. Entry list and odds are readable contract state before you commit. Randomness comes from a Chainlink VRF coordinator we do not control, verified on chain before the callback can pay anybody. The prize sits in the contract, not a company account. The whole explanation of a win is a contract address, published rules, a VRF request, a proof and a block number. No second layer where an interested party estimates a value.
This is not a claim that a raffle beats a credit fund; they do different jobs and one of them funds actual commerce. It is a claim about why one can be fully verified and the other cannot. A draw has nothing off chain to be wrong about. Attach an asset with an uncertain future cash flow and the problem is back for good.
The honest limits
VRF does not price a receivable, and no block explorer will tell you whether a shop in Jakarta pays in November. Provable fairness is a narrow guarantee about a draw, not a truth machine.
It cuts at us too. A prize denominated in a stablecoin inherits that stablecoin’s off-chain half, the argument we ran in August when the BIS went after par redeemability. Tether’s figures look strong, roughly $1.5 billion in net operating profit last quarter against $187.8 billion in assets and a $4.11 billion buffer, and every one of them is an attestation rather than a proof. And because Base is public, our own entries and payouts are permanently readable by anyone. That is the price of verifiable gaming, printed rather than buried behind a fairness page.
The question worth asking
Not whether StableFund is good or bad. It is probably fine and quite possibly good. Ask the question that works on everything: which half of this is a proof, which half is a conclusion, and who gets paid on the conclusion. The ledger underneath can be perfect and still tell you nothing about what is written on top of it.
📷 Photo by Nick Hillier on Unsplash
