Blockbooster: Interpreting the Limitations and Possibilities of On-Chain Native Credit Origination

Original author: @BlazingKevin_, Blockbooster Researcher

On December 11 of last year, a16z crypto released its annual “Big Ideas 2026: Part 3.” In the stablecoin section authored by partner Sam Broner, the following content is worth discussing:

“Stablecoins that lack robust credit infrastructure resemble narrow banks—holding only specific liquid assets deemed extremely safe. A narrow bank is an effective product, but I don’t believe it will be the long-term backbone of the on-chain economy.”

Broner then offers his judgment:

“We have seen a new batch of asset managers, curators, and protocols beginning to facilitate on-chain asset-backed loans supported by off-chain collateral. These loans are typically originated off-chain and then tokenized. I see little benefit from tokenization here... Therefore, debt assets should be originated on-chain, not tokenized after being originated off-chain.”

Four months later, in March 2026, Sam Broner left a16z and founded The Better Money Company. a16z crypto led a $10 million seed round, with Circle co-founder Sean Neville also participating. However—what Broner himself embarked on was not the ‘on-chain native credit origination’ he pointed out in his article, but another track: a stablecoin clearinghouse, facilitating low-cost swaps between different compliant stablecoins. Signed partners include issuers and distribution channels like Paxos, Stripe's Bridge, and MoonPay.

The person most optimistic about stablecoin infrastructure and the first to call out the “narrow bank ceiling” chose the clearing/interoperability layer over the credit origination layer when starting his own venture. This is because the credit origination layer is too difficult; no project has matured enough for him or equally significant players to bet their own time on it. In other words, even the person who best understands this judgment is still waiting for the “actionable moment” for the credit layer.

This is the theme we discuss today: While the entire industry talks about ‘RWA tokenization,’ the next real structural opportunity might be ‘on-chain native credit origination’—a direction repeatedly discussed but never scaled by anyone.

0. Defining “On-Chain Native”

“On-chain native credit origination” has two easily confused interpretations; we focus on the second.

The first is ‘on-chain native’ in a procedural sense: The entire process, from loan origination and interest rate pricing to liquidation and disposal, is completed on-chain. In this sense, Aave, Compound, and Morpho are already thoroughly on-chain-native—loans are initiated on-chain, rates are algorithmically priced based on capital utilization, and liquidations are automatically executed by smart contracts when collateral ratios are breached.

The second is ‘on-chain native’ in a credit assessment sense: Underwriting credit using the borrower's on-chain behavior, cash flow, and on-chain identity, rather than relying on over-collateralization with crypto assets or off-chain traditional credit reports and financial statements. This is the truly immature part.

The fundamental difference lies in "what justifies the loan." Aave's model is "over-collateralization"—you must first lock in $150 worth of ETH to borrow $100. This is not credit in essence; it's a pawn shop. It creates no new purchasing power; it merely releases liquidity from existing assets. The borrower must have money first to borrow.

True credit origination is "lending based on the assessment of future repayment ability"—a bank lends you money to buy a house based on your income, credit history, and repayment capacity. This kind of credit creates new purchasing power and is the core engine of the money multiplier and economic growth in the modern economy.

Here, a common misconception needs explanation: "Isn't Aave's algorithmic interest rate a form of on-chain underwriting?" No, it's not. Aave's algorithm prices interest rates based on capital utilization, not borrower risk. The more money borrowed from the pool, the higher the rate—this prices the pool's capital tightness, treating all borrowers the same. Aave gives the same rate to every borrower in the same pool because it does not differentiate who the borrower is. True underwriting, in essence, prices different borrowers differently based on their risk—that is the core of credit origination. A system that does not differentiate borrowers, regardless of algorithmic complexity, is not underwriting.

1. Current State

Regarding this direction, there are products on the market, with 5 to 10 teams seriously attempting it, but their combined TVL is still less than the fraction of a single USDC pool on Aave. For example:

• 3Jane: Is the closest attempt to "on-chain native credit underwriting." It uses zkTLS technology to pull a borrower's off-chain banking data (via Plaid integration) and on-chain asset profile. A proprietary underwriting algorithm called 3CA calculates a real-time "Jane Score" credit score, then grants unsecured USDC credit lines—borrowers need not lock any crypto collateral. Default disposal follows real legal channels: bad debts are packaged and auctioned to US collection agencies, with recovered funds distributed between agencies and lenders.
• It raised a $5.2 million seed round led by Paradigm in June 2025, with participants including Coinbase Ventures, Wintermute, Robot Ventures—Circle co-founder Jeremy Allaire was also an angel investor. 3Jane launched its mainnet in early November 2025, with an initial cap of approximately $50 million, initially only available to US residents with total assets over $150,000.

However, even for this most-watched, Paradigm-backed, Delphi-supported star project, its actual TVL remains extremely small (tens of thousands of USD in early stages).

• Divine Research: Represents a route diametrically opposite to 3Jane. Divine is a San Francisco-based company founded by Diego Estevez. Since December 2024, it has been issuing unsecured USDC short-term loans through a platform called Credit—by the second half of 2025, it had issued over 500,000 loans covering more than 100,000 borrowers and completed a $6.6 million financing round.
• Its underwriting approach is based on identity + progressive establishment of repayment history: Borrowers must first verify their unique identity through Sam Altman's Worldcoin using World ID for iris scanning. They then start with a very small credit line (typically under $100); each time they repay fully, the limit increases, up to around $1000. It mainly targets populations in developing countries (Argentina, Nigeria, Colombia, etc.) overlooked by traditional finance—in the founder's words, "high school teachers, fruit vendors... basically anyone with internet access." Interest rates range from 20% to 30%.
• The default rate for first-time loans is indeed high, around 40%, but as borrowers accumulate records within this "repayment for higher limits" flywheel, the overall default rate has been reported to approach zero—the 40% is the acquisition cost at the very front end (covered by high interest and reclaiming distributed WLD tokens from users), not the steady-state bad debt rate of the model.

Placing 3Jane and Divine side by side reveals two routes for on-chain native credit and their respective limitations:

3Jane follows the "asset/income proof" route—using zkTLS to verify your bank account and on-chain assets, targeting asset-rich borrowers (high-net-worth individuals, businesses), with post-default enforcement relying on US legal debt collection. Its limitation: it serves those who already have assets, remaining distant from the true credit origination of "creating purchasing power for the asset-poor," and legal collection is only effective in mature jurisdictions like the US.

Divine follows the "identity proof + progressive trust" route—first using iris scanning to ensure one person has only one borrowing identity, then using the "repayment for higher limits" flywheel to cultivate credit bit by bit. It targets the long-tail, asset-poor populations in developing countries, truly touching inclusive credit. It has no collateral to recover and no effective cross-border legal recourse; the only consequence of default is "this iris can no longer borrow money"—which sounds like a weak deterrent. However, the near-zero steady-state default rate shows this positive incentive of "to borrow more, you must repay first" actually works for long-tail borrowers. Divine's real limitations lie not in deterrence, but in two points: first, the credit it builds is only valid internally within Divine; second, its entire anti-Sybil mechanism is outsourced to World ID, an off-chain biometric identity, rather than solving the pseudonymity problem natively on-chain.

The contrast between these two routes leads to a conclusion: Neither has solved the core question of "what justifies the loan" in the hardest setting—"on-chain, facing a pseudonymous borrower." Instead, each has introduced an external fulcrum. 3Jane uses "proof you have money" to bypass it (which is essentially disguised collateral); Divine uses World ID to anchor identity and the "repayment for higher limits" progressive flywheel to extract credit from behavior. In other words, the hardest version—"judging whether a borrower you neither know and who can change addresses at will will repay in the future based solely on on-chain behavior"—has not been tackled head-on by either route; their cleverness lies precisely in finding a fulcrum that allows them to lend money without having to solve it directly.

Other players include: Wildcat Finance (on-chain matching for bilateral private credit, where lenders and borrowers negotiate terms directly, and the protocol only acts as a matching engine and smart contract executor, with lenders coordinating collection among themselves upon default); Clearpool, TrueFi (varying attempts at unsecured/low-collateral institutional lending); Union Protocol (credit based on social relationships); Accountable (verifiable credit disclosure for off-chain assets). The TVL of these protocols mostly ranges from hundreds of thousands to a few million dollars, with a few institution-oriented ones being larger.

Here you might wonder: Why are these small teams doing this, while the largest DeFi lending protocols—Aave, Morpho, Compound—don't do unsecured lending themselves? They have the deepest liquidity, strongest brands, and most on-chain data, making them seemingly best positioned for on-chain native underwriting. They don't do it for two structural reasons:

• First, tail risk cannot be borne by token holders: Over-collateralized liquidation is automatic and predictable, while losses from unsecured credit defaults are real bad debts. Governance token holders cannot bear such credit tail risk—a large-scale default could wipe out the entire protocol.
• Second, regulatory arbitrage space: Over-collateralization has a clear "non-security, non-traditional lending" legal narrative (essentially collateral swaps), while unsecured lending would immediately enter the purview of consumer credit regulation. Thus, the business models and risk structures of incumbents determine that they cannot and will not do this—ironically, this creates a structural window that new teams can enter, which giants cannot.

Now, let's answer another question: Where exactly is the demand? If it's merely "theoretically it should exist," then this is a story of finding a solution for a problem. But real on-chain credit demand is already distributed across several concrete scenarios: market makers and quant teams need working capital for turnover but are unwilling to lock up equivalent collateral; on-chain native merchants, RWA asset originators, and Crypto projects need accounts receivable financing and prepayments; and the vast number of small and medium-sized borrowers directly excluded by the over-collateralization model—they lack surplus crypto assets to pledge but have real cash flow.

In other words, the over-collateralization model serves "people who already have money and want to release liquidity." The demand excluded by this model is precisely "people with cash flow but lacking collateral"—this is the real market for credit origination. Demand is screened out by the collateral threshold of the existing model and has never been counted.

2. Why Stablecoins "Need" to Solve This Problem

To understand why on-chain native credit origination is "a structural need," one must first understand the traditional monetary banking concept of "Narrow Bank."

The Narrow Bank is a classic theoretical construct: a bank that only takes deposits, holds only ultra-safe assets (short-term Treasuries, central bank reserves), and makes no loans. A narrow bank's deposits are 100% backed by safe assets, theoretically immune to runs and bankruptcy. It sounds safe, but it has never become mainstream historically because it has a fatal commercial ceiling: it does not create credit, therefore does not generate a money multiplier, and its profit potential is extremely limited.

The core value of modern banks lies precisely in "fractional reserve + credit origination." You deposit $100; the bank keeps a portion as reserves and lends out the rest. The lent-out money becomes someone else's deposit and is lent out again... This process creates purchasing power far exceeding the original deposit (the money multiplier) and is the financial engine of modern economic growth. The narrow bank voluntarily abandons this engine, so it can only be a marginal player in the financial system, not a pillar.

Now look at the stablecoin system—it is a giant narrow bank. USDC, USDT absorb "deposits" (user funds), hold reserves 100% in short-term Treasuries and cash, make no loans, and create no credit. The entire stablecoin market's "deposit" size—approximately $240 billion in mid-2025, surpassing $320 billion by mid-2026—all sits in safe assets, generating no money multiplier.

A misunderstanding must be avoided here: "Not generating a money multiplier" does not equal "not profitable." On the contrary, issuers are extremely profitable—they keep the Treasury interest from the reserves. The GENIUS Act + CLARITY Act prohibits paying interest to holders, not issuers profiting from the spread. Thus, the problem with stablecoins is not that "no one profits," but that: this profit is locked at the issuer layer, neither shared with users nor entering the multiplier cycle of credit creation. Value is captured, not amplified.

Therefore, if the stablecoin system wants to break through the narrow bank ceiling and truly become an "on-chain banking system," the only path is to create credit outside the issuer—that is, at the DeFi protocol layer. And the current credit at the DeFi protocol layer is not true credit origination; it's just a pawn shop.

Thus, the logical loop closes: Stablecoin issuers are legally prohibited from lending → Credit origination can only happen at the protocol layer → The existing over-collateralized model at the protocol layer does not create new purchasing power → So the only logical path for the stablecoin system to break the narrow bank ceiling is to develop genuine on-chain native credit origination.

3. Why Is It Stuck So Far?

If on-chain native credit origination is a structural inevitability, why have only 5-10 teams attempted it for over a year, and why is TVL still not scaling?

The answer is a chicken-and-egg dilemma, but a more accurate historical parallel is the US consumer credit market before FICO.

Engineer Bill Fair and mathematician Earl Isaac founded Fair, Isaac and Company as early as 1956, but the consumer-facing FICO credit score wasn't officially launched until 1989. Widespread industry adoption, becoming a standard for lending, didn't happen until the mid-1990s after the GSEs (Fannie Mae, Freddie Mac) adopted it for mortgages. From company founding to score launch: 33 years. To industry-wide adoption: about 40 years.

The maturation of credit infrastructure layers is measured in "decades," not "years." And it was this FICO score that first made credit calculable, reusable, and standardized across institutions. The decades following FICO's adoption saw the true explosion of the US consumer credit market—the scaling of credit cards, auto loans, and mortgages followed the standardization brought by FICO. FICO is not a feature of consumer credit; it is a prerequisite for its scaling.

What on-chain credit lacks now is precisely this "FICO moment"—a widely accepted, mechanistically credible, and reusable "on-chain credit score."

Without this standardized credit layer, every protocol attempting on-chain native credit is forced to build its underwriting system from scratch: 3Jane builds its own 3CA algorithm and Jane Score; Spectral builds credit scores based on on-chain wallet behavior; Cred Protocol, Blockchain Bureau each have their own on-chain credit models; identity layers include attempts by Worldcoin, Gitcoin Passport. Every protocol reinvents the wheel; no standard is reusable by others. This is like the US before FICO—each banker had their own set of subjective judgments, unable to scale.

All current attempts at on-chain native credit are stuck in a chicken-and-egg loop: true on-chain credit assessment requires rich on-chain credit history, but most real borrowers' economic activities are still off-chain; there isn't enough behavioral data on-chain to support underwriting. Thus, protocols are forced either to fall back on off-chain data or to limit lending to "the wealthy whose assets are already on-chain." Both paths fail to reach the long-tail borrowers who truly need credit creation.

But the FICO analogy also diagnoses a deeper sticking point. FICO's success wasn't just standardizing credit scores, but also standardizing the consequences of default—once you default, your FICO score is visible to the entire industry, affecting your ability to borrow from any institution. This "cross-institutional transmissibility of default consequences" is the real deterrent power of FICO: not punishment from a single bank, but punishment from the entire financial system.

On-chain credit has yet to establish this kind of "cross-protocol transmissibility": default penalties from a single protocol cannot take effect across others, so each protocol's bad debt risk is locked within itself, unable to be diluted through an industry-wide reputation mechanism.

A true "on-chain FICO" must solve both score standardization and cross-protocol transmission of default consequences. Many are attempting the former; almost no one has touched the latter—and the latter is further blocked by the more fundamental issue of "persistent anti-Sybil identity."

4. Interim Solutions

Returning to the current market reality, we believe that infrastructure layer (persistent identity + cross-protocol default broadcasting + standardized scoring) might be very difficult to establish. Therefore, given that the endgame may be unreachable, whoever can bypass it and capture value at an interim stage appears more valuable at the current moment.

First, understand why these three things are difficult—independently so. This means betting "they will all be built" is essentially betting on a series of locks being opened simultaneously:

• The First Lock, Data Pipeline: Technologies like zkTLS bring off-chain data onto the chain credibly—but this ironically proves that there isn't enough credit data on-chain itself; it has to be borrowed from off-chain. A system entirely reliant on bringing bank statements or VantageScore on-chain is "on-chain" only in its transmission encryption; the substance of underwriting remains off-chain credit checks. So the assumption that "data pipelines will become the basis for truly on-chain native underwriting" is fragile.
• The Second Lock, Credit Bureau—This is the most valuable layer but also the least likely to emerge spontaneously, as it's a classic public good/coordination problem. First, see how traditional credit bureaus formed—through decades of industry consolidation, regulatory push, and ultimately oligopolistic mergers. None of the three major bureaus emerged because a startup "built a better protocol." Expecting an open protocol to spontaneously grow a widely adopted credit bureau on-chain within a few years is akin to treating something shaped over half a century through regulation and mergers as a product deliverable by engineering.
• The Third Lock, Persistent Anti-Sybil Identity—This is the most fundamental and possibly fundamentally unsolvable layer. It contains an inherent contradiction: any identity binding strong enough (mandatory KYC, biometrics) sacrifices core on-chain attributes like openness and permissionlessness, turning on-chain credit back into a traditional system requiring centralized identity endorsement. Any solution light enough to preserve permissionlessness is vulnerable to "start over with a new address."

Under the constraints of these three locks, we believe on-chain native credit origination is a direction where the endgame is extremely difficult to reach.

And all the products currently running are, without exception, "bypassing those locks." They each borrow the missing component of the endgame from outside the chain (off-chain law, biometric identity) rather than building it on-chain. So, are there interim bypasses with broader coverage or more "on-chain native" characteristics than these two?

5. Better Interim Directions

Deconstructing the endgame reveals what's stalling everyone is essentially the same thing: The endgame requires "punishment" to be effective—if you default, the consequence must be able to follow you across protocols and addresses. And "effective punishment" precisely relies on those three most difficult things: persistent identity, cross-protocol transmission, credible 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.

Unpack this asymmetry. To "punish" a defaulter, you must make all protocols see their stain—a public goods dilemma, no one wants to supply it. To "reward" a good performer, you only need to give addresses with clean histories better terms within your own protocol. "The cost of rebuilding" is constrained by the reward mechanism.

This flips the entire problem. The endgame pursues "how to ensure defaulters cannot escape"; interim products pursue "how to let compliant actors accumulate something of increasing value." The latter is the form on-chain credit is most likely to succeed in first.

This "reward compliance" logic is already working for Divine. Its "repayment for higher limits" flywheel is essentially "exchanging accumulated good repayment records for better borrowing terms (higher limits)."

So the following interim directions apply the same "reward compliance" logic validated by Divine to scenarios it hasn't yet covered—especially into the main DeFi arena where collateral is abundant and capital efficiency is the real pain point. Their commonality: all are built on "rewarding compliance" rather than "punishing default."

Direction One: Progressive Lowering of Collateral Ratios—Let Reputation Provide a "Discount," Not a "Replacement."

The endgame of on-chain native credit is "zero collateral"—an asymptote difficult to touch. But between "150% over-collateralization" and "zero collateral" lies a whole continuous spectrum, and this spectrum itself is a huge, almost untapped market.

The most natural interim product looks like this: Each time a borrower repays on time or safely closes a position in a protocol, it's recorded in that address's performance history. As clean history accumulates, the protocol gradually relaxes requirements—collateral ratios drop from 150% to 130%, 120%, 110%; interest rates get discounts; limits increase; liquidation gets a buffer. This is precisely the path in the real world from "secured credit card → regular credit card → credit limit increase": you first prove yourself with a deposit, then trade your record for the deposit.

Aave's Efficiency Mode (E-Mode) seems somewhat similar. But E-Mode adjusts asset correlation (e.g., between stablecoins, ETH and stETH), not borrower history: it gives uniform treatment to all, looking only at what you've pledged, not who you are or how many times you've repaid.

Direction Two: Replace "Judging the Person" with "Intercepting Cash Flow"

The endgame is difficult largely because it tries to solve one of the thorniest problems: predicting a borrower's character.

On-chain, programmable cash flow can be automatically intercepted at the smart contract layer. If a borrowing entity's future income is already on-chain (a merchant's sales flow, a protocol's fee share, even a tokenized salary stream), then a loan can be structured so that when income arrives, the contract automatically deducts the loan repayment first, and the remainder goes to the borrower. The lender's "collateral" is that future cash flow, custodied by code and inaccessible to the borrower.

Projects like Goldfinch, Centrifuge, and Maple tokenize off-chain generated receivables—underwriting, due diligence, and collection remain off-chain. The real interim opportunity lies with those cash flows that are generated natively on-chain and can thus be directly intercepted by smart contracts.

Direction Three: Curator Model

Since a standardized, credible underwriting algorithm is unlikely to emerge on-chain in the short term, stop pretending algorithms can solve underwriting. Instead, let those who genuinely have underwriting ability and are willing to put up first-loss capital do the underwriting. This is delegated credit and the curator model: the protocol only provides the rails (settlement, transparency, automated contract execution of agreed terms). Who to lend to and under what conditions is decided by a delegate/curator who provides first-loss capital; they earn the interest spread and also bear the first losses.

It doesn't need a universal on-chain FICO; it replaces that universal scoring layer with "localized trust + first-loss capital." Aave's credit delegation, and the curator/vault models being developed by Maple and Morpho, are early forms of this direction. Value will accrue to good curators—whose vaults have low default rates and stable returns will attract more deposits, itself a slowly emerging form of credit anchored in performance.

However, dialectically speaking, it essentially moves the trust problem one layer up—you don't need to trust the borrower, but you need to trust the curator. It's more like "packaging the human-judgment aspect of off-chain credit with on-chain transparency and automated liquidation."

None of these three interim directions attempts to punish default; they focus on "rewarding compliance"—letting an address's accumulated good history gradually translate into lower collateral ratios, priority over cash flows, curator preference, or various tangible benefits within an ecosystem. Punishment requires industry-wide coordination; rewards only require individual protocols or ecosystems to have the incentive.

Therefore, the more likely path for on-chain native credit origination is: individual protocols, individual ecosystems each deepen the practice of "compliant addresses deserve better terms," making on-chain repayment records gradually valuable in specific scenarios. These scattered, reward-anchored accumulations of credit grow address by address, protocol by protocol, eventually becoming thick enough at some point to resemble true credit.