Onchain Credit's Fastest Growing Category Is Built Wrong
Asset-backed credit is the only onchain credit category that can solve adverse selection. Tokenized fund wrappers don't get it there. Birch Hill's thesis on solving it at the vault layer.
Three Markets, Not One
It is impossible to talk about onchain credit without separating the three things people call by that name.
The first is overcollateralized crypto lending. Aave, Morpho, Compound, Spark. A borrower posts $1.50 of ETH to draw $1.00 of stablecoin. Liquidation is mechanical. The collateral is liquid 24/7. The category has matured into one of the most reliable yield primitives in DeFi, with stablecoin supply rates today sitting in the 3.5% to 7% range. This works. It also does not scale beyond crypto-collateralized risk.
The second is uncollateralized lending. This has been the holy grail of DeFi since 2017 and has consistently failed in its permissive forms, because the protocol layer cannot answer the three questions any credit business actually has to answer: who is the borrower, how do you price their default, and what happens when they don’t pay. The history of this category thus far is a graveyard.
The third is asset-backed credit. Loans against real, identifiable collateral, with offchain legal claims, third-party assessment, and recovery rights. This is the category that’s growing fastest. It is also the only category with a credible answer to the adverse selection problem that has killed every previous attempt at credit onchain.
“Asset-backed credit” as a label is already growing onchain, mostly as a wrapper. Today’s onchain ABC is largely tokenized fund interests: a fund holds the loans, the tokens represent shares in the fund, and the structure of the fund determines whether the adverse selection problem actually gets solved. In most cases, the protocol layer just passes through whatever risk the fund sponsor built in.
ABC is the fastest-growing onchain credit category, but the way it’s growing today, wrapped in tokenized funds, does not actually solve the problem. The fund wrapper inherits the structure of whoever sponsored it. What we’re implementing is the solution at the smart contract layer: assessment, structure, and recovery encoded into the vault itself, not inherited from a fund manager offchain.
Where the Growth Actually Is
The onchain real-world asset category, RWAs across all sub-segments, has gone from $5.6B at the start of 2024 to $25.96B as of June 3, 2026. That is a ~4.6x expansion in 29 months, almost all of it concentrated in the last twelve. The pullback off the May peak is small relative to the trajectory and consistent with normal category volatility.
It is worth putting that growth in context. Even after a 4.6x expansion in 29 months, the entire onchain RWA market today is a rounding error against what the market believes the category can become. Deloitte’s April 2025 forecast projects $4 trillion of tokenized real estate alone by 2035, with tokenized loans and securitizations representing the single largest subsegment at $2.39 trillion.
The composition matters more than the headline. The first wave was tokenized U.S. Treasuries: BlackRock’s BUIDL, Franklin Templeton’s BENJI, Ondo, Superstate, and now JPMorgan’s MONY and BNY Mellon’s competing products. These are essentially money-market funds with a blockchain wrapper. They proved that institutions can and will custody onchain, but they did not solve a credit problem. They simply ported an existing product.
The second wave, the one we’re in, is tokenized private credit. Onchain private credit has grown ~180% year-over-year and is now the largest non-stablecoin RWA sector. Maple Finance has $3.17B in its USDC lending vault and ~$926M in its USDT vault, roughly $4.1B in stablecoin deposits across the two. Centrifuge has passed $1.38B in TVL. New entrants keep coming, including Centrifuge’s USDS vault at ~$865M.
These are credit funds with onchain plumbing. The borrowers are KYC’d, the loans carry offchain legal documentation, third-party assessment is done by humans with credit committees, and recovery happens in court like every other private-credit fund. What blockchains contribute is the distribution, the transparency of the wrapper, and increasingly the regulatory containment layer.
Stablecoins are the demand side, and they are growing even faster.
Total stablecoin supply has reached roughly $323B today. USDT is around $190B across chains. USDC is ~$73B. The other yield-bearing and crypto-backed stables (USDS, DAI, USDe, etc.) add another ~$65B. The aggregate float looking for compliant onchain yield is, by our read, well north of $250B and growing at high single-digit percent per month. The question is no longer whether there’s onchain capital that wants yield. The question is what that capital is allowed to do.
What Asset-Backed Credit Actually Is, and Why It Works
When we say asset-backed credit, we mean something specific. A loan is asset-backed when (a) the borrower pledges identifiable real-world collateral, (b) the lender holds a perfected security interest in that collateral, (c) recovery does not require the borrower’s cooperation, and (d) the loan-to-value, term structure, and recovery waterfall are documented in legal instruments that survive default. In traditional private credit, this is unremarkable. In onchain credit, it is the single feature that separates a working structure from a failed one.
The collateral types that work in ABC are the ones with three properties: they are identifiable (not commingled), they have an active secondary market for liquidation, and the lien process is well-understood in the jurisdiction of the borrower. Examples: trade receivables, equipment finance, real-estate-backed bridge loans, invoice factoring, structured consumer finance. Examples that do not work, or work only at much wider spreads: future cash flows of operating businesses without specific asset pledge, intangible IP, illiquid private equity stakes. The discipline is collateral first, borrower second.
Real-estate-backed lending is worth a moment of focus, because it is the single largest projected segment of the entire tokenization opportunity. Deloitte’s breakdown is unambiguous on this point: by 2035, tokenized loans and securitizations are projected to dominate the tokenized real estate mix, at roughly three times the size of tokenized private real estate funds and dwarfing tokenized equity in undeveloped or under-construction projects.
What earns the spread is the work done before deployment. Each loan goes through a credit assessment that prices the probability of default, the loss given default, and the recovery process. Each loan is documented with covenants, reporting obligations, and trigger events. Each pool has a curator or assessment partner with reputational and financial skin in the game. Each vault has a defined waterfall: who gets paid first when assets are liquidated, what triggers a pool freeze, how losses are allocated. The mechanism is doing the work; the marketing is doing nothing.
This is also why the spread persists. As tokenized T-bills cluster at ~3.5%, and overcollateralized DeFi clusters between 2.5% and 4.2%, well-structured ABC pools sit consistently in the 4-5% range with multi-billion-dollar capacity. That premium is compensation for credit work an expert must complete. The protocol layer doesn't do that work, it is built so outside experts can curate it.
The Adverse Selection Problem, and Why Most Onchain Credit Hasn’t Solved It
Anyone who has worked in private credit knows the borrower self-selecting into your product tells you almost everything. The cheapest capital goes to the strongest borrowers. The credit you ultimately get is the credit that was rejected by everyone earlier in the stack. This is the adverse selection problem, and it is the only problem in private credit that actually matters.
Onchain credit, in its permissive forms, has historically inverted the solution to this problem. Early protocols extended capital to pseudonymous wallets willing to pay the highest rate. No identity. No recourse. No assessment layer beyond pool stats. The 2022 stress period made the consequences obvious: pools that had marketed themselves as “institutional-grade” discovered their borrowers were correlated, undisclosed, and in some cases insolvent. Recovery on defaulted loans was effectively zero because the loans were unsecured and the jurisdictions unclear.
The version that works, permissioned pools with KYC, offchain documentation, dedicated curators, and a real assessment layer, is now the dominant structure of the surviving onchain credit market. The onchain market leaders have all converged on essentially the same answer: the credit work happens offchain, the pool encodes the result. That is a meaningful improvement on permissionless undercollateralized lending. It is not the end of the design space.
Here is the key point that today’s market has not internalized: when ABC sits inside a fund wrapper and the fund determines underwriting, recovery, and waterfall, the adverse selection problem moves up one level. If the fund manager has incentive misalignment, leverage exposure, or inadequate assessment, the onchain vault is just a more transparent way to deliver bad risk.
The Regulatory Inflection: GENIUS, CLARITY, and the Yield Question
If the market structure is the first leg of why we are building now, U.S. regulation is the second, and it has moved in the last twelve months in a way most operators are still catching up to.
The GENIUS Act, the first federal stablecoin framework, was signed into law on July 18, 2025. It established a federal regime for payment stablecoin issuers, required 1:1 reserves, and, critically for our thesis, prohibited stablecoin issuers from paying any form of interest or yield directly to holders. The OCC issued its proposed implementing rulemaking on February 25, 2026. The statutory deadline for final regulations is July 18, 2026. The full operational regime is expected to be in place by January 2027 at the latest.
The GENIUS Act left one loophole: it banned issuer-paid yield but didn’t fully address yield paid by exchanges or affiliated platforms on stablecoin balances. The Digital Asset Market CLARITY Act, currently moving through the Senate after passing the House, closes that gap. The latest draft prohibits offering yield directly or indirectly on stablecoin balances. The White House’s April 2026 paper, “Effects of Stablecoin Yield Prohibition on Bank Lending,” makes the policy logic clear: Washington wants stablecoins to be a payments rail, not a deposit substitute, and it wants yield to flow through clearly demarcated investment products rather than through the stablecoin itself.
This is the regulatory window most of the market is mispricing. The set of products that can legally deliver yield to onchain dollars is being explicitly narrowed. The set of structures that survive (registered funds, properly disclosed lending vaults, tokenized credit products) is being implicitly elevated. The next twelve months will be a sorting event.
The Vault Is the Architecture That Matters
If stablecoin issuers cannot pay yield, and exchanges cannot pay yield on stablecoin balances, then the only legitimate way to convert onchain dollars into income is through a discrete, identifiable investment product. Onchain, that product is a vault. ERC-4626 and its successors have become the de facto standard for tokenized yield-bearing positions, and the vault is now the chassis for almost every compliant onchain credit product in the U.S. market.
In traditional asset-backed lending, the wrapper is incidental. You raise an LP commitment, draw down into a fund, deploy into loans, distribute.
Onchain, the vault is doing far more work than its TradFi equivalent. The vault is, simultaneously, the issuance mechanism (it mints shares representing claims on the underlying loans), the disclosure mechanism (its accounting is publicly verifiable), the distribution mechanism (anyone with a wallet can interact subject to permissioning), the recovery mechanism (waterfalls and trigger events are encoded), and increasingly the regulatory containment vessel (permissioning can enforce KYC, accredited-investor gates, and jurisdictional restrictions in ways a stablecoin cannot). When yield is prohibited at the stablecoin layer, the vault layer becomes the place where compliant, regulated, transparent yield gets delivered.
This is why we believe the design of the onchain vault, its permissioning, its accounting, its disclosure standards, its compliance posture, and most importantly the way it encodes credit work, becomes the single most important architectural choice in this market over the next eighteen months. In traditional ABC, you can have a mediocre fund wrapper and great credit work and still produce a great product. Onchain, in the post-GENIUS/CLARITY world, a poorly designed vault becomes a regulatory liability. A vault that simply tokenizes a fund interest relocates the adverse selection problem instead of solving it.
Where We Stand
We focus on asset-backed credit because it is the only onchain credit category with a structurally defensible answer to adverse selection, and because the data is unambiguous about the demand side.
We orient toward the United States with a compliance-first posture because we believe the U.S. regulatory framework, post-GENIUS and post-CLARITY, will define the global standard for how onchain yield is delivered. The cost of retrofitting compliance onto a non-compliant structure is far greater than the cost of designing it in from day one.
We treat the vault layer as a first-order design problem, not a packaging exercise, because we believe the architecture is where the next eighteen months of regulatory, technical, and operational complexity will compound. We are not interested in being another tokenized fund interest. We are interested in encoding assessment, structure, and recovery into the vault itself, at the protocol layer, in a way that the current generation of products does not.
The Next Twelve Months
By July 2026, the OCC’s final GENIUS Act rules will be in place. By the end of 2026, the CLARITY Act framework will likely be enacted or close to it. By January 2027, the full payment stablecoin regime will be operational. Somewhere in that window, a generation of yield-bearing products designed for the pre-2025 regulatory environment will need to retire, and a generation of products designed for the post-CLARITY environment will need to be ready.
The onchain capital base is there. Stablecoin float has reached ~$323B. Onchain credit demand is growing at triple-digit rates. The regulatory framework is being finalized and, with appropriate caveats, pointing in a direction that favors well-structured, compliant, U.S.-domiciled, vault-based asset-backed credit.
We see a rare alignment of market timing, regulation, and capital formation. Birch Hill is building for it.
Disclaimer: This note reflects Birch Hill's house view as of June 3, 2026. It is not investment advice and does not constitute an offer to sell or a solicitation of an offer to buy any security or investment product. Regulatory references describe public legislation and rulemaking and are not legal advice. All third-party data is sourced from publicly available analytics platforms, including DefiLlama, and is current as of the publication date. Past performance is not indicative of future results. Birch Hill is pre-registration as of this date.