The Curator Economy

Why this matters

DeFi lending was supposed to be code, not people. The early thesis was that smart contracts could replace credit officers: collateral ratios, liquidation logic, and interest rate curves would make humans unnecessary. That thesis held for the first cycle of Aave and Compound, when every supported asset was a top-20 crypto with a deep oracle market and a fixed parameter set.

It does not hold now. Onchain credit has expanded into RWAs, restaking derivatives, illiquid LP tokens, stablecoin variants with different reserve compositions, and yield-bearing collaterals where the underlying yield itself can fail. None of these can be safely managed with a fixed-parameter, system-wide risk framework. Someone has to make a call on each one. That someone is the risk curator, and the curator role is now the asset manager class of DeFi.

This page covers what the curator economy is, who runs it today, how institutions are entering, and where the picture being sold by industry boosters diverges from what the on-chain data actually says.

From pooled to curated: the architectural shift

Aave and Compound built single-pool lending. Every asset went into one giant balance sheet. If one collateral type failed (a depeg, an oracle attack, a hyper-volatile token), losses propagated across the entire system. Governance had to defensively manage parameters for the worst-case asset because every asset was systemically connected to every other.

This bundling forced a single risk standard onto a market that contained many distinct risk profiles. The result: conservative pools turned away interesting collateral; permissive pools accumulated tail risk that nobody had explicitly priced. Risk management at that stage was a system-wide tuning exercise, not asset-by-asset judgment.

Morpho Blue, launched in early 2024, broke the bundle. The protocol became a thin, immutable primitive: anyone can spin up an isolated market with a chosen collateral, loan asset, oracle, liquidation LTV, and interest rate model. Losses in one market do not touch another. Risk decisions migrated from protocol governance to the entity that creates the market. (For the architectural details, see Modular Lending & Morpho Explained and Monolithic, Modular, and CeDeFi.)

On top of the primitive sit MetaMorpho vaults: aggregators that let a depositor put a single asset (say USDC) into a vault that then allocates across multiple Morpho Blue markets. The vault has a curator, and the curator chooses which markets to enter, in what proportion, with what caps. The depositor is buying the curator's judgment, not the protocol's. The protocol just executes.

That is the moment the curator economy started. The infrastructure became neutral. The judgment became a service.

Where curators sit in the stack

The DeFi lending stack now maps cleanly onto the division of labor in traditional asset management:

• Capital source: Depositors (retail or institutional). In TradFi this is the limited partner, the pension fund, or the retail investor buying a fund.
• Distributor: The user-facing surface. Coinbase, Kraken, fintech apps. In TradFi this is the brokerage or RIA channel.
• Asset manager: The risk curator. Designs the vault strategy, sets collateral eligibility, decides allocations and caps. In TradFi this is BlackRock, PIMCO, or a credit hedge fund.
• Custody and execution: Morpho Blue itself. Holds the assets, runs the liquidation logic, executes interest accrual. In TradFi this is the custodian bank and the clearing infrastructure.

The mapping is not a metaphor. It is structural. Each layer can be operated by a different entity with different incentives and different regulatory exposure. That separation is what makes the management layer interesting as an institutional entry point: a TradFi firm can show up as a curator without having to build smart contracts, integrate with exchanges, or hold crypto on its own balance sheet.

Who manages the money today

Two views are useful here. The first is Morpho-only, because Morpho Blue is where the curator role first crystallized and where the most granular per-vault data exists. The second is multi-chain, because the same teams have been steadily expanding to Solana, Base, BNB Chain, and multiple lending protocols, and a Morpho-only snapshot now meaningfully understates their actual managed AUM.

Morpho-only view (most granular data)

Source: Morpho Blue GraphQL API (blue-api.morpho.org), top 300 MetaMorpho vaults by AUM, pulled 2026-05-19. Long tail beyond 300 adds under $5M.

Cross-protocol view (the real picture)

The same three teams curate vaults on Euler V2, Kamino, and a handful of smaller venues, with substantial concentrations on Base and Solana. DefiLlama tracks each as its own "protocol" and reports per-chain TVL:

Source: DefiLlama protocol pages (defillama.com/protocol/steakhouse-financial, /gauntlet, /sentora), pulled 2026-05-19. Includes all DefiLlama-tracked curator activity across protocols (Morpho, Euler V2, etc.) and chains.

Three short notes on each team:

• Steakhouse Financial: Most vaults (26), highest concentration. Has positioned itself as the default curator for high-grade collateral, especially USDC, USDT, and stablecoins with explicit reserve compositions. Largest single vault is Steakhouse Prime USDC at $465M. Sets the de facto floor for what gets called a "conservative" onchain stablecoin vault.
• Gauntlet: Originated as a quantitative risk advisor to Aave and Compound before pivoting to curator. Runs the most vaults (43) but at smaller average size, suggesting a granular strategy of asset-specific exposure. The team's edge is simulation infrastructure: they model parameter shocks across markets before deploying capital.
• Sentora: Concentrated in just 2 vaults (PYUSD and RLUSD), both stablecoin-collateralized. Vault sizes are large ($280M and $205M respectively), suggesting institutional inflows tied to specific stablecoin partnerships rather than diversified yield seeking.

Three paths institutions are taking in

Three patterns of institutional entry have emerged. They differ in how much control the institution wants, how much of the capability it has in-house, and how much accountability it absorbs.

1. Distribution: outsource curation, keep the customer

A centralized exchange or fintech with a captive user base partners with an external curator and exposes the curator's vaults as a yield product to its own customers. The exchange handles KYC, deposits, fiat ramps, and front-end UX. The curator handles collateral evaluation and risk allocation. Examples publicly reported in industry coverage: Coinbase routing yield through Steakhouse vaults, Kraken routing through Sentora.

Trade-off: speed of entry is high, build cost is low, but reputational risk and accountability for the curator's choices stay with the distributor. If the curator makes a bad collateral call and depositors lose money, the customer sees the exchange's logo, not the curator's. Distribution requires picking the right curator and then living with that pick.

2. Supply: push assets onto the rails and shape the standards

Asset managers that hold real-world assets (private credit, treasuries, money-market positions) supply those assets to onchain markets as collateral or as yield-bearing tokens. The harder version of this strategy goes further: acquire governance tokens in the lending protocol so the supplier has a voice in which assets get whitelisted as "official" collateral.

Apollo's reported acquisition of Morpho governance tokens is the canonical example. The move is not about owning a token. It is about influencing the standards that determine whether Apollo's private credit products clear the bar for use as onchain collateral. The supplier who shapes the rule book wins permanent placement; the supplier who arrives later competes on price inside a rule book somebody else wrote.

3. Operator: become the curator

The most ambitious path. The institution builds (or hires) an in-house risk team and operates its own vaults directly. The institution captures the management fee, sets the strategy, and owns the brand attached to the vault. It also absorbs all the operating risk: a bad allocation decision is now legally and reputationally attributable to the institution itself.

Bitwise's positioning of onchain vaults as "ETF 2.0" framed this path in the abstract for most of 2025. As of May 13, 2026 it is concrete: Bitwise was announced as the curator of Jupiter Lend's Ethena USDe market on Solana, the first time a US-registered crypto asset manager has held curation authority on a major Solana lending product (per Jupiter Lend's announcement, May 13, 2026). The market is structured as a four-party arrangement: Jupiter Lend provides the lending market layer, Bitwise sets the parameters and underwrites the collateral, Ethena supplies the USDG borrow liquidity, and Fluid runs the underlying infrastructure. The Operator path has moved from positioning deck to live capital.

The economics resemble running an ETF: a published strategy, a management fee, daily transparency on holdings. So do the responsibilities: the operator is fully on the hook for performance and operational failures, and that responsibility cannot be outsourced. The next test of this path is whether Bitwise's risk decisions (oracle choice, leverage caps, eligible collateral) hold up under stress, and whether other US-registered asset managers follow with similar curator-of-record arrangements.

How vault risk actually shows up: seven categories

Before getting to the structural risks the industry isn't pricing, here are the concrete technical risks that have been pricing themselves through losses over the past year. Curated vaults are composable products. They aggregate base protocols, oracles, asset issuers, and operator decisions, so risk shows up at every layer. Per Castle Labs (May 2026), seven distinct risk categories cover almost every incident the sector has actually suffered.

1. Smart contract risk (compounds across three layers)

A depositor in any curated vault is exposed to the vault contract itself, the base lending primitive (Morpho Blue, Euler V2, etc.), and every underlying market the allocator routes into. A single position in Gauntlet's USDC Core vault on Morpho touches the vault contract, Morpho Blue, and each Morpho Market it's allocated to, plus the oracle and any wrapped-token contract for the collateral. The risk surface multiplies, not adds. ERC-4626 even introduced its own attack class: the Donation Attack, where an attacker artificially inflates a fresh vault's share-to-asset ratio so that subsequent depositors mint zero shares and their assets accrue to the attacker. Venus protocol on zkSync lost roughly $700K of wUSDM to this in March 2025.

2. Curator risk (incentive misalignment is the structural one)

Curators are usually paid performance fees tied to yield generated, not flat AUM fees. That creates a baked-in incentive to allocate toward higher-yielding markets even when those markets carry higher tail risk. The curator captures the upside via fees; depositors bear the downside via bad debt. The mitigation is curators with meaningful skin in the game (depositing their own balance sheet alongside LPs), tight separation between curator role and sentinel role (Morpho Vaults V2's role-based architecture: Owner, Curator, Allocator, Sentinel), and explicit asset-solvency-first underwriting rather than yield-first allocation. Sentora explicitly frames its model this way; not every curator does.

3. Rehypothecation and contagion

The Stream Finance collapse (November 2025) is the canonical recent case. xUSD was a synthetic stablecoin accepted as collateral across Morpho and Euler vaults; the protocol had $520M in TVL backed by only $160M in actual user deposits, meaning the system was running roughly 3.25x leverage at the protocol level before any individual user looped. When an external fund manager lost $93M, the cascade through curator-managed vaults turned a $93M direct loss into hundreds of millions in ecosystem impact. Historical comparison: Celsius, BlockFi, Voyager, and Gemini in 2022 reused customer deposits as collateral for lending to Three Arrows Capital; combined losses ran into the tens of billions. Same failure shape, different decade.

4. Counterparty risk (perp vaults are special)

Vaults inherit the risk of every external dependency: underlying protocol failures, asset issuer problems (stablecoin depegs, custody issues), oracle provider failures, team operational failures. Perp vaults face an additional category because their counterparty IS the trader. The JELLY incident on Hyperliquid's HLP showed the shape: an attacker opened $4.05M in longs and a $4.1M short at 20x leverage covering ~40% of JELLY's circulating supply, then pumped the spot price across exchanges to liquidate their own short and trap the HLP on the losing side. HLP couldn't clear the position in the open market, so Hyperliquid's team manually delisted JELLY, closing the short for a $700K gain and avoiding a worst-case loss of more than $13M. The precedent: a "decentralized" exchange manually overriding its own smart contracts to protect a vault.

5. Oracle failures (the Stream-Resolv pattern)

Hardcoded oracles protect against false-positive liquidations during small deviations. They become traps during real depegs. Stream Finance again: xUSD oracle feeds stayed pegged to $1 while the asset traded at a 70-80% discount on the open market. Loans backed by xUSD remained "overcollateralized" on-chain even after the backing had evaporated, blocking liquidations that would have contained losses. Bad debt accumulated across every vault with allocation to xUSD-denominated markets. The Resolv exploit (March 2026) was the inverse-arbitrage version: an attacker minted roughly 80M unbacked USR tokens via a compromised signing key with only ~$100K-200K of real collateral, USR and wstUSR depegged on secondary markets, but Morpho and Fluid kept valuing wstUSR near the pre-depeg oracle price of ~$1.13. Traders bought cheap USR on the open market and deposited it at face value to drain the vaults. Result: ~$10M bad debt across 15 Morpho vaults plus ~$21M on Fluid.

6. Liquidity risk (the 100% utilization trap)

When a curator wants to exit a market to honor depositor withdrawals but the underlying lending market is at 100% utilization, there's nothing to withdraw until borrowers repay. The Stream Finance xUSD cascade hit this exact trap: as depositors rushed to exit, remaining liquidity got borrowed against to fund exits, utilization hit 100%, and the door slammed. Curators couldn't pull capital; new depositors weren't arriving; borrowers had no incentive to repay during a panic. The hardcoded oracle made it worse by blocking the liquidation path that would otherwise have unwound the bad positions. Compounds badly with leverage and RWA redemption windows: a 5x leverage position with a T+2 day asset redemption cycle takes up to 10 days to fully exit.

7. Liquidation risk (oracle latency and cascade dynamics)

Two failure modes here. The first is liquidation-as-bug: in March 2026, Aave's CAPO risk oracle priced wstETH at ~1.19 ETH while the market was at ~1.23 ETH. The 3.3% gap triggered the liquidation of 34 fully-collateralized accounts in a single block, generating $27.78M in erroneous liquidations before the error was corrected. The protocol mechanically did exactly what it was told; the oracle just told it wrong. The second is cascade dynamics: a sharp collateral price decline triggers liquidation bots competing to clear positions, which sells into a falling market and accelerates the decline, which liquidates more positions. Morpho's isolated market design contains this within a single market; Aave's pooled design has had to add circuit breakers and stricter listing review (the V4 Spokes architecture) to limit it.

Risks the curator economy does not yet price

Industry coverage tends to frame the curator market as a maturity signal: capital flowing to specialists, division of labor emerging, institutional standards taking shape. That framing is partially correct and dangerously incomplete. Five structural risks are not yet reflected in how the market thinks about itself.

Concentration is fragility, not maturity

86% of curator AUM in three teams means the system has three single points of failure. A governance attack, an internal compromise, a misaligned strategy, or a single oracle-related liquidation cascade at any one of them removes a major share of the entire curator economy. The TradFi equivalent of three asset managers controlling 86% of a market would attract regulatory scrutiny on antitrust grounds. The DeFi equivalent attracts press calling it "specialization."

Crisis stories are told from the curator's side

Industry coverage often cites the October 2025 Gauntlet pool-recovery episode (a sudden $775M inflow that compressed yields, then "normalized within ten days") as proof of curator capability. Read from the depositor side, the same episode is a 10-day yield gap that the existing depositors funded. "Normalized APY" is a story about the curator regaining target yield; it is not a story about the depositors who were diluted to get there. Curator crisis playbooks have an asymmetric incentive: protecting the franchise matters more than protecting any individual depositor cohort, because the depositors can leave but the franchise cannot.

There is no fiduciary duty

This is the structural gap behind the others. A TradFi asset manager who makes a clearly negligent allocation decision is exposed to legal action by the LPs. A DeFi curator is exposed to capital flight. Capital flight is a market discipline, not a legal one, and it works only when the discipline arrives before the catastrophe. Most curator failures will not give depositors time to leave.

Regulatory perimeter is unclear

A curator setting collateral standards, choosing allocation across markets, charging a fee, and accepting capital from the public looks like an unregistered investment adviser in most jurisdictions. The fact that the curator role is encoded as a smart-contract permission does not make the activity legally distinct from advisory work. The pricing of this risk is currently zero. The first enforcement action will reprice it suddenly.

The same curator runs many vaults

A depositor in two different Steakhouse vaults is exposed to the same operating team, the same risk philosophy, the same key-management infrastructure, and the same potential conflicts of interest. The vault-level isolation that Morpho Blue provides at the smart-contract level does not extend to the curator-level operating risk. "Diversifying across vaults" without diversifying across curators is not actually diversification. (See Vault Credit Risk for the metrics framework on what to evaluate at the vault level.)

Why the standards get set now, not later

The case for paying attention to a $2.91B market when the global asset management industry is $147T sounds backwards. It is not.

Curators are doing the work of writing the playbook for how onchain credit gets underwritten. Which oracles are acceptable. Which RWAs count as conservative. Which liquidation thresholds get used for which collateral types. Which yield-bearing assets are eligible. The curators on the field today are the ones whose conventions become the defaults. The next wave of larger entrants (the BlackRocks, the Apollos at full size, the Fidelitys) will arrive into a market where these defaults are already set, and they will be operating inside someone else's framework.

For institutional observers, the practical implication is that the window for setting standards is narrow and closing. The window for capturing AUM is wide open and growing. These are not the same window, and they reward different actions. Setting standards now requires curation expertise applied at scale. Capturing AUM later requires only distribution.

Investment considerations

For depositors

Three questions to ask of any curator-managed vault before depositing:

• Curator-level diversification: How much of your total onchain exposure runs through this one team? Vault-level isolation is not curator-level isolation.
• Collateral transparency: Can you see, in real time, what the vault is currently allocated to, in what proportion, with what caps? If not, you are buying judgment without monitoring it.
• Liquidation behavior: What happens to your share value if one of the vault's underlying markets gets liquidated? Some vaults absorb losses pro-rata; some have caps that limit exposure to any single market. The difference matters in a stress event.

For curators and would-be curators

The barrier to entering as a new curator is low. The barrier to capturing AUM is high. The capital is following two signals: a verifiable risk-management track record (Steakhouse's stablecoin selectivity, Gauntlet's parameter simulation history), and distribution access (the existing top three each have some form of CEX or fintech routing). A new curator needs at least one of these, and ideally both, to gather meaningful capital.

For TradFi observers

The management layer is where existing TradFi capability is most directly portable. Distribution is already crowded by crypto-native firms. Custody and execution require blockchain engineering that incumbents do not have. Management is the underwriting and allocation work TradFi has done for decades, simply applied onchain. The fastest entry is through Path 1 (Distribution) by partnering with an existing top-three curator. The most defensible entry is through Path 3 (Operator) by building an in-house curator team. Path 2 (Supply) is only sensible for institutions that already hold real-world assets they want priced as onchain collateral.

April 2026 update: yield reversal and the Treasury-parity question

The picture sharpened materially in April 2026. Two data points from the Glassnode April 2026 institutional monthly update (citing James Aitchison, CIO of Caerus Global Management, and Glassnode's own curator vault performance dashboard) deserve to be surfaced because they change the framing of curator-economy investability from "outperforming alternative" to "structural credit allocation, not a substitute."

1. The 3-month USD curator ARR is now below Treasuries

Source: Glassnode April 2026 monthly institutional update (median basis; methodology shifted from average to median in this release to reduce outlier distortion).

After several months of marginal outperformance, the USD curator 3-month annualized return (3.13%) is now below the US Treasury rate (3.78%). The 12-month ARR lead has narrowed to just 37 basis points (4.49% vs 4.12%), the tightest gap of the year. On the ETH side, the curator-versus-staking gap has compressed to just 2 basis points monthly performance and 50 bps on the 3M ARR (down from 73 bps in March). In short: onchain credit no longer pays an obvious premium over the risk-free / native-staking alternatives, and on shorter horizons it now underperforms.

2. Curator yields carry 3-4x the volatility of native benchmarks

USD curator volatility (0.33%) is 3.7x Treasury volatility (0.09%). ETH curator volatility (0.10%) is 3.3x staking volatility (0.03%). The premium that curator strategies historically earned was compensation for genuine risk being absorbed, not free yield. When the monthly performance spread compresses to 2 bps (April 2026 ETH) while the volatility multiple stays at 3-4x, the risk-adjusted return profile deteriorates sharply.

3. Total curator TVL has plateaued, not recovered

Per the Glassnode April 2026 monthly: total curator TVL has stabilized in the $5-6B range through April, well below late-2025 peaks near $10B. This corroborates the cross-protocol $5.14B figure cited earlier on this page (DefiLlama, 2026-05-19) and confirms it is not a recovering number. The absence of meaningful inflows despite tighter monthly performance suggests allocators are holding back on fresh capital deployment until the yield trajectory becomes clearer. Some of this is post-Kelp-DAO-exploit caution: the April Kelp DAO incident impaired or froze a small number of vaults that had held affected collateral, and several positions remain frozen pending governance resolution.

4. The framing has shifted: structural allocation, not Treasury substitute

The bullish framing of the curator economy through late 2025 was approximately "outperformance over Treasury or staking yield with active risk management." The Aitchison/Glassnode April 2026 framing is meaningfully different: "curator exposure as a structural allocation to on-chain credit rather than a Treasury or staking substitute, with sizing calibrated to the higher vol profile". That is a more honest framing and a more constrained allocation case. It says: curators are not the obviously-better-yield trade today; they are a different exposure (credit, smart-contract, oracle risk) that should be sized as such, not as an outperformance bet.

Key takeaways

• Curators are the asset manager class of DeFi. Morpho Blue's modular architecture split risk judgment from execution, and specialist teams now run vaults the way TradFi asset managers run funds.
• The curator economy is bigger and more cross-chain than Morpho-only data suggests. Top-3 curators manage ~$5.14B in total across Ethereum, Solana, Base, BNB Chain, and multiple lending protocols (per DefiLlama, 2026-05-19); Morpho Blue alone carries $2.91B of that. Headline "$7B curator market" figures conflate Morpho Blue total TVL with curator-managed AUM; the more useful number is the cross-protocol curator total.
• RETIRED FRAMING (May 2026): curators are not a Treasury substitute. The 2025 framing of "curator yields outperform Treasury or staking with active risk management" no longer matches the data. Per Glassnode April 2026: USD curator 3M ARR (3.13%) is below Treasuries (3.78%), ETH curator-vs-staking gap is just 2 bps monthly, total curator TVL plateaued at $5-6B (vs $10B late-2025 peak), and curator volatility runs 3-4x native benchmarks. The honest 2026 frame is structural onchain-credit allocation with sizing calibrated to the higher vol profile, not an outperformance bet. TI surfaces have been updated to retire the substitute framing.
• Three institutional entry paths exist, and the Operator path is now concrete. Distribution (outsource curation, keep customers), Supply (push assets onto rails, shape standards), and Operator (build the curator) all have live examples. Bitwise's May 13, 2026 curation of Jupiter Lend's Ethena USDe market on Solana is the first US-registered asset manager operating directly as a curator on a major Solana lending product.
• Not every "curated market" has a third-party curator. Kamino's Ethena USDe market is protocol-self-curated, structurally different from the external-curator pattern Morpho popularized. Both can be called "curated lending," but the accountability shape and entry-path implications differ.
• The accountability scaffolding has not caught up to the structure. No fiduciary duty, no registered-adviser perimeter, no supervised conduct, and a concentration profile (within Morpho 86% in top-3; cross-chain still highly concentrated) that would attract regulatory attention in TradFi.
• The standards window is closing faster than the AUM window. Curators on the field now are writing the rules of onchain credit. Capital arriving later operates inside those rules rather than setting them. The standards-setting argument remains intact even as the alpha-generation argument weakens.