Last December 11, a16z crypto published Big Ideas 2026: Part 3. In the stablecoin section written by partner Sam Broner, one passage is worth dwelling on:
“A stablecoin without robust credit infrastructure looks like a narrow bank — holding only specific liquid assets considered to be extremely safe. The narrow bank is a useful product, but I don’t believe it will be the long-term backbone of the onchain economy.”
Broner then makes his call:
“We’ve already seen a new wave of asset managers, curators, and protocols beginning to facilitate onchain asset-backed loans collateralized by offchain assets. These loans are typically originated offchain and then tokenized. I think tokenization adds almost nothing here … so debt assets should be originated onchain, not originated offchain and then tokenized.”
Four months later, in March 2026, Sam Broner left a16z and founded The Better Money Company. a16z crypto led the $10M seed round; Circle co-founder Sean Neville also participated. But here’s the twist — what Broner went out and built himself is not the “onchain-native credit origination” he named in his own essay. It’s a different lane entirely: a stablecoin clearinghouse, doing low-cost swaps between different compliant stablecoins. Its signed partners are issuers and distribution channels like Paxos, Stripe’s Bridge, and MoonPay.
The person most bullish on stablecoin infrastructure — the one who first called out the “narrow bank ceiling” — chose, when it came time to put his own time on the line, the clearing/interoperability layer rather than the credit-origination layer. Why? Because credit origination is too hard to build, and no project is mature enough to make him, or anyone of comparable caliber, willing to bet their own time on it. Put differently: even the person who understands this thesis best is still waiting for the credit layer’s “ready-to-build moment.”
That’s our subject today: while the entire industry talks about “RWA tokenization,” the next genuinely structural opportunity may be “onchain-native credit origination” — a direction discussed over and over, yet one no one has built to scale.
1. Defining “Onchain-Native”
“Onchain-native credit origination” admits two easily-conflated readings. We’re discussing the second.
The first is process-sense “onchain-native”: from loan origination to rate pricing to liquidation, the entire flow happens onchain. In this sense, Aave, Compound, and Morpho are already thoroughly onchain-native — loans originate onchain, rates are priced algorithmically off utilization, and liquidation executes automatically via smart contract when the collateral ratio breaks.
The second is underwriting-sense “onchain-native”: underwriting credit from a borrower’s onchain behavior, cash flows, and onchain identity — rather than relying on over-collateralization, and rather than relying on offchain credit reports and financial statements. This is the part that genuinely hasn’t matured.
The fundamental difference between the two is on what basis you lend. Aave’s model is over-collateralization — to borrow $100, you must first post $150 of ETH. This isn’t credit; it’s a pawnshop. It creates no new purchasing power; it merely releases the liquidity locked in assets you already own. The borrower has to be rich before they can borrow.
Real credit origination is “lending against a judgment about future ability to repay” — the bank lends you money to buy a house based on your income, credit history, and capacity to pay. This kind of credit creates new purchasing power; it is the core engine of the money multiplier and economic growth in the modern economy.
A common misconception needs clearing up here: “Isn’t Aave’s algorithmic rate a form of onchain underwriting?” No. Aave’s algorithm prices the rate that arises from utilization, not the borrower’s risk. The more of the pool gets borrowed, the higher the rate climbs — that prices how tight the pool’s capital is, identically for every borrower. Aave gives every borrower in the same pool the same rate, because it doesn’t distinguish who the borrower is. Real underwriting, by contrast, is fundamentally about giving different borrowers different prices for different risk — and that is the core of credit origination. A system that doesn’t distinguish between borrowers is not underwriting, no matter how sophisticated its rate algorithm.
2. The State of Play
On this front, products exist, and 5–10 teams are seriously trying — but their combined TVL is still less than a rounding error on Aave’s single USDC pool. For instance:
-
3Jane is the attempt that comes closest to “onchain-native credit underwriting.” It uses zkTLS to pull a borrower’s offchain bank data (via a Plaid integration) and onchain asset profile, runs an underwriting algorithm called 3CA to compute a “Jane Score” credit score in real time, and then issues an uncollateralized USDC credit line — the borrower posts no crypto collateral. Default resolution runs through a real legal chain: bad debt is packaged and auctioned to U.S. collection agencies, with recovered funds split between the agency and the lenders.
-
Its $5.2M seed round in June 2025 was led by Paradigm, with participation from Coinbase Ventures, Wintermute, and Robot Ventures — Circle co-founder Jeremy Allaire is also an angel investor. 3Jane launched mainnet in early November 2025 with an initial cap of roughly $50M, and was initially open only to U.S. residents with total assets above $150,000.
But even this — the most-watched project in the category, backed by Paradigm, endorsed by Delphi — has a tiny actual TVL (on the order of a few hundred thousand dollars early on).
-
Divine Research represents a route diametrically opposite to 3Jane’s. Divine is a San Francisco company founded by Diego Estevez; since December 2024 it has issued uncollateralized short-term USDC loans through a platform called Credit — by the second half of 2025 it had cumulatively issued over 500,000 loans across more than 100,000 borrowers, and raised $6.6M
-
Its underwriting method is identity + progressive build-up of repayment history: a borrower must first pass an iris scan via Sam Altman’s Worldcoin to anchor a unique World ID, then start from a very small line (typically under $100); each time they pay off a loan, the limit ratchets up, to around $1,000 at the top. It serves mainly the populations ignored by traditional finance in developing countries (Argentina, Nigeria, Colombia, etc.) — in the founder’s own words, “high-school teachers, fruit vendors … basically anyone who can get online.” Interest runs 20–30%.
-
Its first-loan default rate does run as high as ~40%, but as borrowers accumulate a record inside this “repay-to-raise” flywheel, the overall default rate has been reported as approaching zero — the 40% is the cost of the very front-end customer-acquisition layer (covered by high interest and the WLD tokens recovered users can claim), not the steady-state loss rate of the model.
Put 3Jane and Divine side by side and you can see the two routes of onchain-native credit, and the limits of each:
3Jane takes the “proof of assets/income” route — using zkTLS to verify your bank account and onchain assets, serving asset-rich borrowers (high-net-worth individuals, businesses), with defaults running through the U.S. debt-collection legal chain. Its limit: it serves people who already have assets, leaving it a distance from the true credit origination that “creates purchasing power for the asset-poor,” and its legal collection only works in mature jurisdictions like the U.S.
Divine takes the “proof of identity + progressive trust” route — first using an iris scan to ensure one person can hold only one borrowing identity, then using a “repay-to-raise” flywheel to cultivate credit bit by bit, serving the long-tail, asset-poor populations of the developing world, genuinely reaching inclusive credit. It does have no collateral to recover and no effective cross-border legal recourse; the only consequence of default is “this iris won’t be able to borrow again” — which sounds like weak deterrence, yet the near-zero steady-state default rate shows that this “want-more-borrow-must-repay-first” positive incentive does in fact work on long-tail borrowers. Divine’s real limit isn’t on the deterrence side; it’s two things: first, the credit it builds is only valid inside Divine; second, its entire Sybil resistance is outsourced to World ID, an offchain biometric identity, rather than solving the pseudonymity problem onchain-natively.
The contrast between these two routes points to one conclusion: neither has solved “on what basis do you lend” under the hardest setting — onchain, facing a pseudonymous borrower. Each instead imports a fulcrum from outside that setting. 3Jane sidesteps with “prove you have money” (which is essentially collateral in disguise); Divine anchors identity with World ID and then squeezes credit out of behavior via the progressive “repay-to-raise” flywheel. In other words, the hardest version — judging, from onchain behavior alone, whether a borrower you neither know nor can pin down (they can swap addresses at any time) will repay in the future — neither route has met head-on. Their cleverness lies precisely in each finding a fulcrum that lets them lend without solving it head-on.
Other players include: Wildcat Finance (onchain matching of bilateral private credit — borrower and lender negotiate terms directly, the protocol provides only the matching engine and smart-contract execution, and on default lenders coordinate recourse among themselves); Clearpool and TrueFi (varying degrees of uncollateralized/under-collateralized institutional lending); Union Protocol (credit based on social relationships); and Accountable (verifiable credit disclosure of offchain assets). Most of these protocols’ TVL sits in the hundreds-of-thousands to single-digit-millions range, with the institution-oriented ones somewhat larger.
You might ask here: why is it these small teams doing this, while the largest DeFi lending protocols — Aave, Morpho, Compound — don’t do uncollateralized credit themselves? They have the deepest liquidity, the strongest brands, the most onchain data; by rights they’re best positioned to do onchain-native underwriting. They don’t, for two structural reasons:
-
First, tail risk can’t be borne by token holders. Over-collateralized liquidation is automatic and predictable, whereas uncollateralized default losses are real bad debt. Governance token holders cannot bear this kind of credit tail risk — one large-scale default could blow through the entire protocol.
-
Second, regulatory arbitrage. Over-collateralization has a clean “not-a-security, not-traditional-lending” legal narrative (it’s essentially a collateral swap), whereas uncollateralized credit immediately enters the field of view of consumer-credit regulation. So it’s the incumbents’ business model and risk structure that make them unable and unwilling to do this — which conversely hands new teams a structural window the giants can’t enter.
Then a further question: where exactly is the demand? If it’s only “there theoretically should be,” then this is a story of inventing a problem to fit a solution. But real onchain credit demand is in fact already distributed across several concrete scenarios: market makers and quant teams need working-capital turnover but don’t want to lock up equal collateral for it; onchain-native merchants, RWA originators, and crypto projects need receivables financing and advances; and a large set of small-to-medium borrowers shut out directly by the over-collateralization model — they have no spare crypto to pledge, but they do have real cash flow.
Put differently: the over-collateralization model serves “people who already have money wanting to release liquidity,” while the demand locked out — the “has cash flow, lacks collateral” cohort — is precisely the real market for credit origination. That demand was filtered out by the collateral threshold of existing models, and was never counted in the first place.
3. Why Stablecoins “Need” to Solve This
To understand why onchain-native credit origination is a structural need, you first have to understand the classic monetary-banking concept of the “narrow bank.”
The narrow bank is a textbook theoretical construct: a bank that only takes deposits, only holds ultra-safe assets (short-term Treasuries, central-bank reserves), and makes no loans at all. Its deposits are 100% backed by safe assets; in theory it can never face a run, never go bankrupt. It sounds very safe, but it has never become mainstream historically — because it has a fatal commercial ceiling: it creates no credit, therefore produces no money multiplier, and its profit potential is extremely limited.
The core value of the modern bank lies precisely in “fractional reserves + credit creation.” You deposit $100; the bank keeps a fraction as reserves and lends out the rest; the lent money becomes someone else’s deposit and is lent out again … this process creates purchasing power far beyond the original deposit (the money multiplier), and is exactly the financial engine of modern economic growth. The narrow bank deliberately gives up this engine, which is why it can only ever be a peripheral player in the financial system, never a backbone.
Whether onchain credit creation can truly produce a money multiplier hinges on one precondition — whether the lent stablecoins can be redeposited into the protocol and become a fresh source of lending. If they can (akin to the supply→borrow→re-supply loop on Aave), then a multiplier-like effect does emerge; if borrowers mostly take the funds offchain to spend, with the money leaving the onchain credit system, then the multiplier effect is limited. So strictly speaking, onchain credit creation is a necessary condition for a money multiplier, but how far the multiplier can amplify still depends on the capital-recirculation rate of the onchain economy.
Now look at the stablecoin system — it is itself a giant narrow bank. USDC and USDT take in “deposits,” hold reserves that are 100% short-term Treasuries and cash, make no loans, and create no credit. The entire stablecoin market’s “deposit” base — roughly $240B in mid-2025, surpassing $320B by mid-2026 — sits entirely in safe assets, producing no money multiplier at all.
One misreading to avoid: “produces no money multiplier” does not mean “makes no money.” Quite the opposite — issuers are extremely profitable; they keep the Treasury yield on the reserves. The GENIUS Act + CLARITY Act prohibit paying interest to holders, not the issuer earning the spread itself. So the stablecoin problem isn’t “nobody profits from it”; it’s that this profit is locked at the issuer layer — neither shared with users nor fed into the multiplier loop of credit creation. Value is intercepted, not amplified.
Therefore, if the stablecoin system wants to break through the narrow-bank ceiling and truly become an “onchain banking system,” the only way out is to create credit outside the issuer — that is, at the DeFi protocol layer. And the credit at the DeFi protocol layer today isn’t real credit creation; it’s a pawnshop.
So the logical loop closes: stablecoin issuers are legally barred from lending → credit creation can only happen at the protocol layer → the protocol layer’s existing over-collateralization model creates no new purchasing power → therefore the only logical way for the stablecoin system to break the narrow-bank ceiling is to develop genuine onchain-native credit origination.
4. Why Is It Still Stuck?
If onchain-native credit origination is a structural inevitability, why have only 5–10 teams tried it for over a year, and why is TVL still failing to scale?
The answer is a chicken-and-egg dilemma — but a more precise historical analogue than “chicken-and-egg” is the U.S. consumer-credit market before FICO.
Engineer Bill Fair and mathematician Earl Isaac founded Fair, Isaac and Company back in 1956, but the consumer-facing FICO score wasn’t formally launched until 1989, and it wasn’t truly accepted industry-wide as the lending standard until the mid-1990s, after the mortgage GSEs (Fannie Mae, Freddie Mac) adopted it. From company founding to score’s birth was 33 years; to industry-wide adoption, roughly 40.
Maturation of the credit-infrastructure layer is measured in decades, not years. And it was precisely this FICO score that, for the first time, made credit something computable, reusable, and standardizable across institutions. In the decades after FICO spread, the U.S. consumer-credit market truly exploded — the scaling of credit cards, auto loans, and mortgages all followed in the footsteps of FICO standardization. FICO isn’t a feature of consumer credit; it is the precondition for consumer credit’s scaling.
What onchain credit lacks right now is precisely this “FICO moment” — a widely-accepted, mechanically-credible, cross-protocol-reusable “onchain credit score.”
Without this standardized credit layer, every protocol doing onchain-native credit is forced to build its underwriting system from scratch: 3Jane builds its own 3CA algorithm and Jane Score; Spectral builds a credit score off onchain wallet behavior; Cred Protocol and Blockchain Bureau each build their own onchain credit models; and at the identity layer, Worldcoin and Gitcoin Passport are taking their own swings. Every protocol is reinventing the wheel, and no single standard is reusable by the others. It’s like the pre-FICO U.S. — every banker had his own set of subjective judgments, and none of it could scale.
Every current attempt at onchain-native credit is stuck in a chicken-and-egg loop: real onchain credit assessment needs rich onchain credit history, but most real borrowers’ economic activity is still offchain, so there isn’t enough onchain behavioral data to support underwriting. So protocols are forced either to fall back on offchain data, or to restrict lending to “the rich whose assets are already onchain.” Neither route reaches the long-tail borrowers who actually need credit creation.
But the FICO analogy diagnoses a deeper sticking point. FICO succeeded not only because it standardized credit scoring, but because it also standardized the consequence of default — once you default, your FICO score is visible industry-wide, and your future borrowing at any institution is affected. This “cross-institution conductivity of default consequences” is the true source of FICO’s deterrence: it’s not one bank punishing you, it’s the entire financial system punishing you together.
Onchain credit has yet to build this “cross-protocol conductivity”: a single protocol’s default penalty can’t take effect across protocols, so each protocol’s bad-debt risk is locked inside its own body, unable to be diluted through an industry-level reputation mechanism.
A true “onchain FICO” must solve both score standardization and cross-protocol conductivity of default consequences. Many are already attempting the former; almost no one has touched the latter — and the latter is in turn stuck on an even more fundamental layer: durable Sybil-resistant identity.
5. Stopgap Solutions
Returning to the current market: we believe that infrastructure layer (durable identity + cross-protocol default broadcast + standardized scoring) may be very hard to build at all. So, given that the endgame is unreachable, whoever can route around it and capture a stretch of stopgap value looks more valuable at this point in time.
First, why are these three things hard — and independently hard? This matters, because betting that “they’ll all get built” is essentially betting on a string of locks all being opened at once:
-
Lock one, the data pipeline. What zkTLS-type tech does is bring offchain data onchain in a trustworthy way — but that precisely proves the reverse: onchain doesn’t have enough credit data of its own, so it has to borrow from offchain. A system that fully depends on porting up bank statements and VantageScore has “onchain” merely adding a layer of encryption to the transmission; the substance of the underwriting is still offchain credit reporting. So the assumption that “the data pipeline will be built into an onchain-native underwriting foundation” is itself fragile.
-
Lock two, the credit bureau — the most valuable of the three layers, and also the least likely to emerge spontaneously, because it’s a classic public-good / coordination problem. Look first at how traditional credit bureaus formed: through decades of industry consolidation, regulatory push, and eventual oligopolistic M&A. Not one of the big three emerged because some startup “built a better protocol.” Expecting onchain to spontaneously grow a widely-integrated credit bureau via an open protocol within a few years is to treat something that took half a century of regulation and M&A to forge as a product that can be engineered and shipped.
-
Lock three, durable Sybil-resistant identity — the most fundamental layer, and the one most likely to be fundamentally unsolvable. It contains a deadlock: any identity binding strong enough (mandatory KYC, biometrics) sacrifices the openness and permissionlessness that are onchain’s core properties, turning onchain credit back into a traditional system requiring centralized identity backing; while any scheme light enough to preserve permissionlessness can’t stop “just spin up a new address and start over.”
Under the constraint of these three locks, we consider onchain-native credit origination a direction whose endgame is extremely hard to reach.
And the products already running, without exception, are all “routing around those locks.” They’ve all borrowed the missing endgame component from outside the chain (offchain law, biometric identity), rather than manufacturing that component onchain. So: is there a stopgap detour with broader coverage, or one more “onchain-native,” than these two?
6. Better Stopgap Directions
Pry the endgame open and you find that what stalls everyone is, at bottom, the same thing: the endgame requires “punishment” to take effect — when you default, that consequence has to be able to chase you across protocols and across addresses. And “punishment taking effect” depends precisely on those three hardest things: durable identity, cross-protocol conductivity, trustworthy data.
“Punishment” is a public good — no one has the incentive to build it alone; but “reward” is a private good — every protocol has the incentive to build it.
Unfold this asymmetry. If you want to punish a defaulter, you have to make all protocols see his stain — that’s the public-goods dilemma, and no one wants to supply it. But if you want to reward a good performer, you only need to grant some benefit, within your own protocol, to addresses that have built a clean history. “Rebuilding has a cost” is enforced by the reward mechanism.
This flips the whole problem over. The endgame chases “how to make defaulters unable to escape”; the stopgap product chases “how to let those who honor their obligations accumulate something increasingly valuable.” The latter is the form onchain credit is most likely to get working first.
This “reward good behavior” logic — Divine has already gotten it working. Its “repay-to-raise” flywheel is, in essence, “using accumulated good repayment records to exchange for better borrowing terms (a higher limit).”
So the stopgap directions below take the same “reward good behavior” logic Divine has validated and carry it into the scenarios it hasn’t yet covered — especially into DeFi’s main battlefield, where collateral is abundant and capital efficiency is the real pain point. Their common thread: all are built on “rewarding good behavior” rather than “punishing default.”
Direction One: progressively lowering the collateral ratio — let reputation act as a discount, not a substitute.
The endgame of onchain-native credit is “zero collateral” — an asymptote, very hard to actually touch. But between “150% over-collateralized” and “zero collateral” lies an entire continuous spectrum, and that spectrum is itself a vast, almost untouched market.
The most natural stopgap product looks like this: every on-time repayment and every safe close-out a borrower makes on a protocol gets logged into that address’s performance file; as clean history accumulates, the protocol gradually relaxes its requirements — collateral ratio drops from 150% to 130%, 120%, 110%, the rate gets a discount, the limit rises, liquidation gets a little buffer. This is exactly the real-world path of “secured credit card → ordinary credit card → limit increase”: you first prove yourself with a deposit, then use your record to shed the deposit.
Aave’s efficiency mode (E-Mode) looks somewhat similar. But E-Mode tunes asset correlation (e.g., between stablecoins, between ETH and stETH), not the borrower’s history: it treats everyone identically, looking only at what you pledge, not at who you are or how many times you’ve repaid.
Direction Two: swap “judging the person” for “intercepting the cash flow.”
The endgame is hard largely because it has to solve a most thorny problem: predicting a borrower’s character.
Onchain, programmable cash flow can be automatically intercepted at the smart-contract layer. If a borrowing entity’s future income is itself onchain (an onchain merchant’s sales flow, a protocol’s fee share, even a tokenized salary stream), then the loan can be designed so that when income comes in, the contract automatically deducts repayment first, and only the remainder goes to the borrower. The lender’s “collateral” is that future cash flow — escrowed by code, untouchable even by the borrower.
Goldfinch, Centrifuge, and Maple and their ilk do the work of bringing offchain-generated receivables onchain — underwriting, due diligence, and collection are all still offchain. The real stopgap opportunity is the cash flows whose income itself occurs onchain and can therefore be intercepted directly by the contract.
Direction Three: the curator model.
Since onchain won’t grow a standardized, trustworthy underwriting algorithm in the short term, then stop pretending you can solve underwriting with an algorithm; instead, let the people who actually have underwriting ability and are willing to eat the first loss themselves do the underwriting. This is delegated credit and the curator model: the protocol provides only the rails (settlement, transparency, contract-automated enforcement of agreed terms); who to lend to, and on what terms, is decided by a delegate/curator who deposits first-loss capital, earns the spread, and is also first to lose. It doesn’t need a universal onchain FICO; it replaces that universal scoring layer with “local trust + first-loss capital.” Aave’s credit delegation, and the curator/vault models Maple and Morpho are building, are all early forms of this direction. Its value accretes to good curators — whoever’s vault doesn’t blow up over the long run and returns steadily attracts more and more deposits, which is itself a slowly-grown, performance-anchored form of credit.
But dialectically: it essentially moves the trust problem up one layer — you don’t have to trust the borrower, but you do have to trust the curator. It’s more like “wrapping the human-judgment element of offchain credit in onchain transparency and automatic liquidation.”
And these three stopgap directions — not one of them tries to punish default; what they do is reward good behavior — letting an address’s accumulated clean history gradually convert into a lower collateral ratio, priority over cash flow, a curator’s favor, or all sorts of concrete conveniences within the ecosystem. Punishment requires industry-wide coordination; reward requires only that a single protocol, or a single ecosystem, has the incentive.
So the more probable path for onchain-native credit origination is this: protocol by protocol, ecosystem by ecosystem, each builds deep the idea that “an address that honors its obligations deserves better terms,” letting onchain performance records slowly become valuable in one concrete scenario after another; these scattered, reward-anchored accumulations of credit grow up address by address, protocol by protocol, until at some point they thicken into something that starts to look like real credit.
——————————————————————
About BlockBooster：
BlockBooster is a next-era alternative asset management firm for the digital age. The firm leverages blockchain technology to invest in, incubate, and manage the core assets of this new era, from Web3-native projects to real-world assets (RWA). As value co-creators, BlockBooster is dedicated to unlocking the long-term potential of these assets, capturing exceptional value for its partners and investors in the digital economy.
Disclaimer:
This article/blog is provided for informational purposes only. It represents the views of the author(s) and it does not represent the views of Movemaker or its affiliates. It is not intended to provide (i) investment advice or an investment recommendation; (ii) an offer or solicitation to buy, sell, or hold digital assets, or (iii) financial, accounting, legal, or tax advice. Digital asset holdings, including stablecoins and NFTs, involve a high degree of risk, can fluctuate greatly, and can even become worthless. You should carefully consider whether trading or holding digital assets is suitable for you in light of your financial condition.
Please consult your legal/tax/investment professional for questions about your specific circumstances. Information (including market data and statistical information, if any) appearing in this post is for general information purposes only. While all reasonable care has been taken in preparing this data and graphs, no responsibility or liability is accepted for any errors of fact or omission expressed herein.
