3Jane lent itself $128.3 million of USD3 deposits, at a 0% risk premium. In six months, $7.25 million of outside money came back.
USD3 is a stablecoin yield product built on loans to real businesses. We followed the dollars from the deposit button to the last wallet the chain can see.
- $128.3M
- 0%
- $7.25M
- 6.0%
- 93%
- 8.4%
01 · The pitchA stablecoin yield built on real loans
Put dollars into 3Jane, and you get a token called USD3. The token's price rises a little every day. That rise is your yield, and 3Jane's docs show it at about 13.1% per year in their main example.
3Jane is a lending protocol on Ethereum. Most crypto lenders, like Aave, make borrowers lock up more crypto than they take out. 3Jane lends without that, the way a bank does, against a borrower's income and credit. The industry calls this tokenized private credit: loans to real businesses and consumers, funded with stablecoins.
Deposits sit in two layers, called tranches. USD3 is the senior tranche, which gets paid first and loses money last. sUSD3 is the junior tranche: it earns more, takes the first losses and so acts as a cushion for everyone holding USD3.
Bitquery reads every transfer on Ethereum, so we followed the dollars from USD3 to whoever borrowed them. We read the lending pool, the borrower's wallet and the loan settings 3Jane wrote on-chain from May to November 2026. Below, we show who borrows the money, what it pays, what has come back and what USD3 has really earned. The trail ends at a cash-out wallet. Past that point, the loans are off-chain, and we rely on 3Jane's own figures, which Yearn's risk team published on 9 November.
02 · The designHow 3Jane says USD3 works
3Jane's docs describe two kinds of loans. The first was credit lines for crypto users in the US, based on their wallet, bank account and credit score. The docs now say those lines are being wound down, and existing borrowers can only repay.
The second kind funds other lenders. Fintech companies that lend to small businesses and consumers need money to make more loans. 3Jane supplies it in two ways.
In a warehouse loan, 3Jane lends against a pool of the fintech's loans, and the fintech puts in its own money too, money it loses first if its borrowers default. In a forward flow agreement, 3Jane buys the loans outright as the fintech makes them. The fintech keeps no stake in those loans. It only promises to take back loans that break the agreed rules.
The docs explain the safety of USD3 with the warehouse. Their example is a $10 million line to LendSwift, a consumer lender that takes the first 25% of any loss. On top of that sits the junior tranche. The docs run a stress test on this setup and find that USD3 only loses money in a crisis on the scale of 2008.
03 · The borrowerWhere 3Jane's USD3 deposits went
The pool has one big borrower. It is a Safe, a shared wallet that only acts when its owners sign. This Safe has a single owner: 3Jane's own multisig, the 3-of-5 wallet that runs the protocol.
Between May and November, the Safe drew $128.3 million from the pool in 31 draws. Almost every time, it sent the money to a relay wallet within minutes, and the relay wallet passed it on to a large cash-out wallet. The exception was the first big draw in May, which sat in Aave for 11 days. The cash-out wallet has sent $3.3 billion to Coinbase Prime, an exchange desk for institutions. That pattern fits a bank or an over-the-counter desk, a broker for large trades. 3Jane says it moves dollars through Erebor, a US bank for crypto firms. We could not confirm that this wallet belongs to Erebor.
So the lending pool does not lend to many fintechs. It lends to 3Jane. 3Jane then lends the dollars out again off-chain and reports what it bought.
You can follow every address in this story with the Bitquery MCP.
04 · The rate3Jane set a 0% risk premium on its own loan
Every borrower on 3Jane has three settings, which 3Jane writes into its smart contracts after a 24-hour delay. There is a credit limit and a value of the assets behind the loan. There is also a risk premium, the extra interest a borrower pays on top of the pool's base rate.
3Jane charged retail borrowers real premiums. Three credit lines set in May carried between 2.6% and 4.0% per year.
3Jane's own Safe paid nothing extra. 3Jane set the Safe's credit line 9 times between May and November. The premium was 0% every time.
The asset value tells a similar story. From late June on, 3Jane set it to exactly 2.857 times the credit limit, so the loan always came to 35% of the assets behind it. A real valuation would move as loans are made and repaid. This one moved only when the limit did, because it was worked out from the limit.
With no premium, the Safe pays two things. The first is Aave's interest on USDC, because the pool lends a wrapped Aave token that grows at that rate. The second is the pool's base rate, the rate every borrower pays, which was between 0.7% and 1.1% per year in the records we read. A zero premium can make sense if the Safe is just 3Jane's funding account and the loan interest reaches depositors in another way. USD3's yield suggests some of it does.
05 · The loansWhat a forward flow agreement means for USD3
According to 3Jane, the money bought loans from three fintechs. Slope makes credit lines for small businesses, Settle offers business financing too, and LendSwift makes small loans to consumers.
Most of the money did not go into the structure the docs use as their example. The two forward flow agreements, with Slope and Settle, hold 93% of the money deployed. In those deals, no fintech takes the first losses. If those borrowers stop paying, 3Jane and then its depositors take the hit.
The docs mention Slope once, as an $8.5 million loan purchase, while the balance 3Jane reported to Yearn is about 7 times that. Settle does not appear in the docs at all. The docs' stress test still holds for LendSwift. It says little about the rest of the book.
Slope has strong backers: J.P. Morgan led its 2024 funding round. We found nothing that says these loans are bad. The problem is that depositors cannot see them, and the safety net in the docs does not cover most of them.
06 · The cashMoney out, and money back
Most of these loans run for less than a year, and borrowers usually repay in monthly installments, so you would expect cash to flow back to the pool steadily.
It has not. Over six months, two payments came back.
The first, in August, came from the cash-out wallet and paid down $7.25 million. The second payment was a round trip. The Safe drew $7,486,914 late on 29 October, and the money went to the cash-out wallet. The next evening, the same amount came back to 3Jane's multisig, which repaid the pool. On the pool's books, that counts as a repayment.
There is an ordinary reading of this. In a revolving deal, a lender can reuse the money borrowers repay to buy new loans, off-chain. Then the pool never sees the cash. But it means depositors rely on 3Jane's reports for everything past the cash-out wallet, including whether the loans pay at all.
07 · The yieldWhat USD3 has earned so far
The real test of a stablecoin yield is the token's price. We read the USD3 price from the deposits and withdrawals people made. It grew 6.0% per year between mid-May and mid-November.
A plain USDC deposit on Aave paid 3.6% per year on average over the same months. So USD3 holders earned about 2.4 points more than Aave, for taking on loans to real businesses. That is about half the figure in 3Jane's docs.
The rates written on-chain do not explain all of that. With no premium on the Safe, the pool should earn roughly Aave's rate plus its small base rate, about 4.5% per year by our rough estimate. USD3 earned more, so 3Jane must be adding interest to the pool some other way. Yearn's report says the protocol's multisig pushes it in directly, and that $8.6 million of booked profit has not yet reached the price.
That works, but it puts the size and timing of the yield in 3Jane's hands. On-chain, most of the yield on the lent money is interest the Safe owes, added to its debt. Only the idle cash, about 11% of the pool, earns interest that arrives as cash. USD3 holders also earn JANE, 3Jane's token, which they cannot sell yet.
08 · The riskWho takes the loss if the loans go bad
If the loans lose money, the insurance fund pays first. It holds about $1.04 million, and 3Jane has not added to it or drawn on it. Next come sUSD3 holders, and only after both are gone do USD3 holders lose money. 3Jane also decides when a loss counts: a bad loan hits the USD3 price only when its multisig marks it, Yearn found.
On 18 November, the cushion covered 8.4% of the loan book. A bigger loss reaches USD3 holders. The docs use a 15% junior tranche as their example.
The cushion is also thin in a second way. Two holders own 74% of sUSD3. One is a wallet that borrows USDC on Aave against USDe. The other is Pendle, where traders split yield-bearing tokens into a fixed part and a variable part.
Pendle matters for USD3 too, because it holds seven in ten USD3. The Pendle market listed in 3Jane's docs ends on 17 December, when traders who hold the fixed part can swap it back. USD3 also serves as collateral, the deposit that backs a loan, on Morpho, where Yearn counts $53 million borrowed against USD3 and its Pendle version. When the Pendle market ends, some of those holders may want dollars back. The pool had $14.0 million of idle cash on 18 November. We have seen a similar unwind before, when USDe loops came apart.
The deposit history shows how quickly money can leave. Withdrawals beat deposits every month from January to April this year. Then June brought the biggest net inflow since USD3 launched, in the same month that the Safe drew $31.2 million.
09 · For holdersWhat it means if you hold USD3
If you hold USD3, you are lending to 3Jane. Your safety rests on loans you cannot see, on a cushion under 10% of the book, and on 3Jane's own choice of when a loss counts and how much interest reaches you.
So far, nothing has gone wrong on-chain. Yearn found no borrower marked as defaulted and no loss posted to the USD3 price. The yield is real, and it has beaten Aave's.
Three things are worth watching. Check whether cash starts to come back to the pool. Check what happens to the idle cash around 17 December. And check whether 3Jane publishes, for every deal, the loan-level data its docs describe.
10 · The limitsWhat the chain cannot show
Past the cash-out wallet, the money is off-chain. The split between Slope, Settle and LendSwift comes from 3Jane, as Yearn published it, and no outside firm has checked those balances. We could not either.
A 0% premium and few repayments could fit a normal setup, where 3Jane runs the loans through its own entity and passes the interest on as it chooses. USD3's extra yield and Yearn's $8.6 million of booked profit fit that reading. The pool's records do not show how much the loans really earn, and 3Jane's docs do not describe this setup.
We did not name the cash-out wallet. Its links to Coinbase Prime come from Bitquery's own labels.
11 · The methodHow we traced USD3
We started at the lending pool and listed every borrower. We followed the money from the biggest one, 3Jane's Safe, wallet by wallet, until it reached a cash-out wallet. We read the loan settings straight from the transactions 3Jane sent. We worked out the USD3 price from real deposits and withdrawals, the holder split from token balances, and Aave's average rate from DeFiLlama's daily record. Wallet names come from Bitquery's own labels.
Bitquery also sells the full Ethereum history as files, as the Ethereum dataset on the Bitquery Data Store. For cases like this, we also run crypto investigation services. For background on lenders that ran out of borrowers, see our piece on dead DeFi lenders.
Ethereum data comes from Bitquery and was read on 18 November 2026. Draws and repayments after that date are not covered.
The loan split between Slope, Settle and LendSwift, the $8.6 million of booked profit and the Morpho borrowing come from Yearn's 9 November report, which rests on 3Jane's own dashboard. Nobody outside 3Jane has checked the loan balances.
Ownership of the Safe and of 3Jane's multisig comes from the public Safe service, read on 18 November 2026.
Aave's average USDC rate is the mean of DeFiLlama's daily readings for the Aave USDC market on Ethereum. The rate moves within each day, so a single reading can differ a lot from the average.
Our estimate of what the on-chain rates alone would pay is rough. It uses the two base-rate readings we took, and the base rate moves with demand.
Draws count transfers of $10,000 or more from Aave to the Safe. We left out $13.2 million that the Safe parked in Aave in May and took back out.
The cash-out wallets are unlabeled. Their links to Coinbase Prime and Galaxy Digital come from Bitquery's labels. We do not say who runs them.
The USD3 price on 5 September comes from Yearn. The other prices come from deposits and withdrawals on-chain. JANE rewards are not valued.
The pool's loans equal its assets minus its idle cash. A second estimate, from the interest the pool logged on 17 November, gives the same figure.
3Jane's docs say the dollars move through Erebor, a US bank. We did not confirm that the cash-out wallet belongs to Erebor.
12 · The recordThe addresses behind USD3
Follow the money in plain English
Every draw, repayment and holder in this story comes from Bitquery's Ethereum data. The Bitquery MCP server puts that data behind an AI assistant, so you can ask where a wallet's money went without writing a query. The full history is also sold as files, as the Ethereum dataset on the Bitquery Data Store.
This article is provided for informational and educational purposes only and reflects analysis of publicly available on-chain data and published documents as of the dates indicated. It does not constitute legal, financial, compliance, or investment advice, and nothing in it is a recommendation to buy, sell or hold USD3, sUSD3, JANE or any other asset.
The findings describe transactions, balances and settings observed on Ethereum between August 2025 and 18 November 2026. Attributions come from on-chain behaviour, public Safe records and Bitquery's own labels; they may be incomplete or incorrect.
References to 3Jane, Yearn, Aave, Morpho, Pendle, Slope, Settle, LendSwift, Erebor, Coinbase, Galaxy Digital, J.P. Morgan or any other named project, company or service describe what public records show. They do not allege wrongdoing by any of them.
Nothing herein should be relied upon as a definitive determination of fact. Readers should conduct their own independent verification before taking any action. The authors and publisher accept no liability for any loss or damage arising from reliance on this content.
Reported by Gaurav Agarwal for Bitquery Research, with AI tools; every figure was checked against the raw data.
FAQ
What is 3Jane?
3Jane is a lending protocol on Ethereum. It lends without crypto collateral, against a borrower's income and credit. Depositors fund the loans through two tokens, USD3 and sUSD3. 3Jane calls this a credit-based money market.
What is USD3?
USD3 is 3Jane's senior deposit token. You deposit USDC and get USD3, and its price rises as the loans earn interest. USD3 gets paid first and loses money last. sUSD3 is the junior token, which earns more and takes the first losses.
Who borrows the money in USD3?
One Safe owes almost all the loans. It is a shared wallet, and its only owner is 3Jane's multisig. That Safe drew $128.3 million in six months and sent it to a cash-out wallet. 3Jane says the money bought loans from Slope, Settle and LendSwift.
How much yield does USD3 pay?
The price of USD3 grew 6.0% per year from mid-May to mid-November 2026. Aave paid 3.6% on plain USDC over the same months, and 3Jane's docs show about 13.1% in their main example. Holders also earn JANE rewards, which cannot be sold yet.
Is 3Jane safe?
Nothing has failed on-chain, and Yearn found no loss recorded. But USD3 depends on loans that only 3Jane can see. The first-loss cushion was 8.4% of the loans in November, and most loans have no lender sharing the losses.
What is a forward flow agreement?
It is a deal to buy a lender's new loans as the lender makes them. The buyer owns the loans outright. The lender keeps no stake in them and only promises to take back loans that break the agreed rules. Slope and Settle sell loans to 3Jane this way.
What happens on 17 December?
A Pendle market for USD3 ends that day. Pendle holds seven in ten USD3. Some holders may want their dollars back then, and USD3 can only pay out from the pool's idle cash.
Run the next investigation in minutes
Ask any question of the blockchain. Bitquery's AI Investigations agent traces funds, attributes wallets, and writes auditable reports across every major chain. Prefer to do it yourself? Connect the Bitquery MCP server to your AI assistant, pull the raw datasets, or query the chain directly with the Bitquery API.