Executive Summary
In early DeFi, most users were effectively their own asset managers, rotating between lending markets and DEX pools with little shared infrastructure to automate rebalancing or compare risk across strategies. First‑generation yield aggregators like Yearn and auto-compounders like Beefy turned that manual workflow into a robo-advisor concept where users would deposit once, and let the vault contract allocate capital toward the best opportunities, often judged on pure APY alone.
Today, onchain asset management is a broader ~$20B category that includes onchain allocators, yield aggregators, and other professionally-run (by seasoned risk managers, market makers, asset issuers, etc) curated vaults that sit on top of DeFi’s growing number of yield sources. Lending markets, DEXs, points programs, RWAs, and structured products all offer different risk-reward profiles that vary across timeframes.
Filtering by “highest” APY is no guarantee of expected returns, as the composition and reward logic of onchain strategies is becoming increasingly complicated and path-dependent. Given the complexity associated with strategy underwriting, most users are better off outsourcing operations to specialized risk curators and asset managers that can price risk and rebalance portfolios on their behalf with 24/7 monitoring.

Since Morpho Blue launched in Jan 2024, curator-led vaults have grown significantly as other lending markets (Euler v2) and yield aggregator vaults (Veda, IPOR) implemented a similar unopinionated agnostic model for asset and risk managers to curate strategies on top, with 30+ teams like Gauntlet, Steakhouse, RE7, Sentora acting as protocol risk managers, vault curators, or strategy managers.

In parallel, a new class of TradFi‑backed onchain vaults has appeared, where regulated RWA products like tokenized Treasuries and money market funds (BUIDL‑style funds, USYC, etc.) are wrapped into tokens and plugged directly into the same DeFi rails. Tokenized Treasuries alone now sit at roughly $13B in onchain AUM and make up the single largest segment of the tokenized RWA market.
Key Takeaways
- Vaults are no longer just APY routers - Onchain asset management has moved from auto-compounding to risk-managed allocation.
- The bottleneck moved from execution to risk selection: Gas optimization and auto-compounding are no longer the main edge. The hard questions are now collateral risk, oracle risk, liquidity risk, maturity risk, and whether incentive-driven yields are durable.
- ERC-4626 lowered integration costs - Standardized vault interfaces made yield products easier to plug into wallets, exchanges, and other protocols.
- The asset management stack is split into layers - Yield substrates, risk managers, infrastructure, strategy curators, and distributors now perform distinct roles.
- Distribution is becoming the moat - CEXs, wallets, and fintech apps can route user deposits into vaults without exposing the DeFi plumbing. What users see is a simple interface while the products abstract the underwriting and infrastructure underneath.
Yield Aggregation to Asset Management
DeFi Summer gave birth to yield farming in 2020, with returns coming from a small set of core venues in growth stages to bootstrap liquidity (meaning huge APYs from token issuance and subsidies, rather than organic activity). This included protocols like Aave and Compound for lending, Curve for stablecoins, Synthetix for synthetic assets, the Uniswap vs Sushiswap DEX wars for the AMM side, and others. TVL went from ~$700M to ~$15B in under a year.

In the early days, to gain market share, protocols bootstrapped liquidity by paying users in governance tokens. Lenders earned incentives, and borrowers were often paid to borrow (enough to offset interest rate payments, and more). On DEXs, LPs received emissions to offset impermanent loss, encouraging them to supply for prolonged periods of time. With so many protocols in this race, capturing the opportunity was a tedious process of monitoring rates, moving capital between venues, claiming rewards, selling, re-depositing, compounding, and repeating whenever incentives shifted.
The first generation of vaults solved the operational problem. Andre Cronje built Yearn to automate the rebalancing work active farmers were already doing across those lending venues. It was a protocol he built to solve his own treasury management problem as the key person running operations for the Fantom Foundation. Beefy did something similar for AMM LP positions, autocompounding DEX LP rewards so users did not have to claim, sell, and restake the rewards (paying gas fees in each step). These models worked because the main inefficiency was execution. Yields were visible, gas was expensive, and the edge came from automating repetitive actions at scale. The vault's value proposition was gas optimization and auto-compounding, not necessarily underwriting.

As DeFi matured, the easy farming edge disappeared. Token incentives became less reliable, gas optimization mattered less on cheaper chains and L2s, and yield sources multiplied. Credit risk, collateral risk, oracle risk, bridge risk, and liquidity risk all became more consequential than harvesting rewards. The category shifted toward asset management.

The modern allocator is outsourcing a portfolio decision. The manager must decide which assets are acceptable collateral, which markets deserve capital, which incentives are sustainable, how much liquidity to keep for withdrawals, and how aggressively to chase yield when spreads widen. As a result, depositors no longer underwrite just the yield source itself. Their capital passes through multiple layers of delegated judgment, from protocol risk managers, vault infrastructure, strategy curators, and distributors, all before reaching the underlying opportunity.

In the old stack, risk concentrated at the protocol level. In the new stack, risk is distributed across market creators, vault infrastructure, curators, distributors, collateral issuers, oracle providers, bridges, and depositors. Yield aggregation is all about finding routes that optimize certain values, like APY; asset management decides whether certain routes are worth taking.
ERC-4626: Standardizing the Vault Interface
ERC‑4626 emerged in 2022 as a formal Ethereum standard to normalize how yield‑bearing vaults work onchain. Before ERC-4626, every lending market, aggregator, and auto-compounder used its own deposit, withdrawal, share-pricing, and redemption logic. Integrators built custom adapters for each vault, raising engineering cost, audit burden, and integration risk. This duplicated work, increased the number of audits, and presented higher smart‑contract risk as the number of strategies and adapters grew to support more assets and protocols.
ERC-4626 standardized the interface for vaults and gave integrators a predictable way to support deposits, withdrawals, redemptions, and vault shares. Integrators only need to support one interface instead of dozens, which reduces development cost, audit burden, and the chance of integration bugs.
A vault can be a simple receipt for one yield source, or it can be an active asset-management vehicle with discretion to route capital across DeFi.
- Protocol-specific vaults/receipt vaults: These accept one asset and expose users to a single protocol, market system, or yield source (e.g. MetaMorpho, Euler Earn, and Aave Earn). Their main function is to standardize accounting, collateral use, and integrations. The relevant question here is what does this receipt represent?
- Cross-protocol asset management vaults: These also use vault shares, but their purpose is to allocate capital across multiple yield sources (e.g. Veda, IPOR Fusion, and Yearn-style allocator vaults). They can route into Aave, Morpho, Pendle, RWAs, credit pools, or other strategy modules. The relevant question here is what discretion does the manager have over my capital?
How the Stack Works
Onchain asset management is best understood as a stack.
At the bottom are yield sources, where returns are generated. Around those venues sit protocol-level risk managers, who define or recommend guardrails. Above them are asset-management infrastructure providers, which provide vault contracts, accounting, routing, and share issuance. On top of that infrastructure sit strategy portfolio managers, who decide how capital should actually be allocated. Finally, the distribution layer packages those strategies into user-facing earn products.

Yield Sources as Substrate
The yield substrate is where returns originate. It includes lending markets, AMMs, rate markets, structured products, tokenized credit, and RWAs.
The most consequential substrate today is lending. Lending lets users earn on a single asset such as USDC, ETH, or BTC without taking impermanent loss from two-sided LP exposure. Borrowers pay interest. Depositors receive that interest, sometimes with additional rewards or points.
Lending now comes in 3 broad forms:
- Pooled protocols like Aave v2/v3 and Compound v2 share one pool across many collateral assets, with governance setting which assets are allowed, at what LTVs, and with what rate curves.
- Isolated designs like Morpho Blue markets, Euler v2 vaults, and Silo split risk into many small markets, each with a fixed collateral asset, borrow asset, oracle, and LTV band, so a bad asset or oracle in one market cannot drain others.

- Hybrid lend-and-liquidity designs such as EulerSwap and Fluid-style systems combine lending and trading liquidity. They let issuers create borrowing markets and trading depth around the same asset without forcing users into conventional two-sided LP positions.

On top of lending sits structured yield and RWA credit protocols, so the effective substrate is the ~$90B DeFi opportunity set across lending markets, structured protocols, and onchain credit platforms where capital can be parked to earn.
The most common underlying yield source, because of its simplicity and customization, is Morpho Blue. Blue is an immutable lending primitive in which Morpho Markets are deployed, not a managed market in itself. Morpho markets are created from 5 fixed inputs: lend asset X, borrow against collateral Y, price it through oracle Z, use interest-rate model R, and cap borrowing at liquidation LTV L. Anyone can create a market without governance approval, and once live, that market configuration is immutable.

MetaMorpho vaults sit above those markets. Depositors put one asset into a vault and receive ERC-4626 shares. The curator decides which Blue markets the vault can allocate to, how much each market can receive, and how withdrawals or reallocations are handled. This turns fragmented immutable markets into usable lending products without making Morpho responsible for market selection.

Euler v2 is built from two primitives: the Euler Vault Kit (EVK), which lets builders deploy customized ERC‑4626 credit vaults, and the Ethereum Vault Connector (EVC), an immutable router between them, allowing collateral posted in one vault to safely back borrowing in another, so an issuer or curator can, for example, combine a lending vault and a DEX vault for the same asset without merging their risk surfaces.
In practice this means some vaults feel paternalistic, with a manager or DAO setting risk limits for everyone, while others are fully permissionless and leave risk selection to users or aggregators. The EVC lets all of those patterns coexist in one modular lending system.

Structured yield and credit sit beside lending. Pendle, for instance, engineers yield by stripping yield‑bearing tokens into fixed‑rate Principal Tokens and variable Yield Tokens so users and curators can take explicit rate views; similarly, IPOR IRD (Interest Rate Derivatives; a primitive separate from IPOR Fusion) runs interest‑rate derivatives that turn floating DeFi yields into more predictable exposures.

Onchain credit platforms like Maple and Centrifuge originate loans against tokenized collateral under pool‑specific rules, then expose pool tokens or wrappers that can be held directly or used inside lending markets. The same capital can move through several of these modules at once. For instance, an Aave v3 USDC position can be stripped into a Pendle PT and rehypothecated as collateral in a Morpho Blue market. The choice to use those building blocks, and in what size, is a portfolio decision.
Curators as Protocol-level Risk Managers
The first curator layer sits close to the underlying yield source. These teams optimize for venue resilience rather than depositor returns, defining what collateral can be accepted, how much leverage is allowed, what supply and borrow caps apply, and how the protocol should respond during stress events. Their goal is to keep a venue solvent and operational across market cycles, even if that means sacrificing some yield.

On pooled venues like Aave, teams such as LlamaRisk analyze collateral behavior and tail scenarios, then set or recommend which assets can be listed, what LTVs and liquidation thresholds they get, supply and borrow caps, and how interest rate curves should respond to utilization. Their mandate is venue‑level solvency and resilience with the goal of avoiding bad debt, limit cascade liquidations, and keep the protocol within a defined risk budget, even if that means passing on some yield.
A similar role, but at a higher level, exists in products like Kraken’s DeFi Earn, where Veda supplies the vault infrastructure, but the three USDC strategies are explicitly risk‑managed by Chaos Labs and Sentora, who choose the underlying DeFi markets, set risk bands, and monitor positions on behalf of Kraken users.

Protocol-level risk managers define the boundaries of a venue. They are not necessarily building a portfolio for the end depositor. Strategy managers like Veda, in contrast, may use the venue, but still have to decide whether the venue’s markets fit their mandate.
Asset Management Infrastructure
Asset-management infrastructure is the operating layer that turns yield sources into managed products. The difference is in discretion, since some vaults allocate within one protocol, while others can route capital across multiple protocols, assets, and strategy modules.
Protocol-specific vaults are constrained allocators. MetaMorpho routes one asset across selected Morpho markets, Euler Earn routes within Euler’s vault system, and Aave Earn vaults deposit into Aave v3 markets and issue ERC-4626 shares. These products improve UX, accounting, and risk segmentation, but the opportunity set is isolated within one venue.
Cross-protocol asset management vaults have broader discretion. Veda, IPOR Fusion, and Yearn-style allocator vaults can sit above multiple yield sources, using adapters or strategy modules to move capital into lending markets, structured yield, credit pools, RWAs, or protocol-specific vaults underneath.

Single-protocol Allocation Vaults
A user deposits one asset and receives one vault share. The vault allocates across multiple markets inside one protocol. The curator decides which markets are approved and how capital is weighted.
MetaMorpho is the canonical example, sitting on top of Morpho Blue. A USDC vault may allocate across cbBTC/USDC, WETH/USDC, sUSDe/USDC, PT-sUSDe/USDC, and other isolated markets, each with its own collateral, oracle, and LLTV. The curator allocates that asset across a set of approved Blue markets. The underlying markets remain immutable; the vault controls allocation.
Euler Earn follows the same pattern inside Euler V2, allocating across approved EVK strategies. Euler's architecture lets collateral, liability, oracle, and liquidation logic compose through the EVC (Ethereum Vault Connector).
Earn vaults are attractive because they are transparent and constrained. Users can inspect current allocations; curator powers are bound by protocol-level rules; and withdrawals work like lending withdrawals: instant when liquidity is available, delayed when capital is fully borrowed. The tradeoff is that this type of single-venue Earn vaults optimize within one lending environment but cannot run unconstrained cross-protocol strategies.
Multi-strategy Vaults
Multi-strategy vaults offer more flexibility. A manager can move capital wherever the best risk-adjusted opportunity appears, and the vault can support themed strategies around ETH staking, BTC yield, stablecoin credit, or ecosystem pre-deposit campaigns. The cost is a larger trust surface as these vaults should be analyzed more like funds.
Users deposit a single asset but the vault can deploy across many protocols, strategies, and sometimes chains. A USDC vault might lend on Aave, allocate to Morpho markets, hold Pendle PTs, provide liquidity, enter a basis trade, or route capital to a RWA pool.
Veda’s BoringVault infrastructure is built for this category of vaults. The ERC-4626 vault holds assets and mints/burns shares, while external modules handle deposits, pricing and execution. Curators are whitelisted with constrained mandates and use it to route a vault’s capital across many yield substrates.

IPOR Fusion approaches the same market through Plasma Vaults, Fuse modules, and offchain strategy logic. The vault is the custody and accounting layer, Fuses are protocol connectors that let the vault interact with external venues, and Alphas execute strategies within the limits set by Atomists (IPOR’s term for curators).

Yearn V3 is also an asset‑management protocol built around allocator vaults and tokenized strategies, all using ERC-4626. Allocator vaults accept deposits and decide how much capital to give each attached strategy, while each tokenized strategy is itself an ERC‑4626 contract that deploys into exactly one underlying yield substrate. Strategists write these single‑source strategies, and the allocator vault rebalances between them over time, so from a user or integrator’s perspective a yVault is one share‑issuing ERC‑4626 asset, but under the hood it is an actively managed basket of protocol positions

To an exchange, wallet, or fintech, these protocols look like one standardized vault API, and to underlying protocols it looks like a normal depositor. This lets multiple frontend distributors and curators share the same vault infrastructure while still being able to target different assets, chains, and risk profiles.
Curators as Strategy Portfolio Managers
The second curator layer sits above the vault infrastructure. These teams behave more like portfolio managers than protocol risk committees, deciding how capital moves across approved yield sources, when to chase yield, when to hold liquidity, and when to reduce exposure.
This creates a new business model around specialized risk management. Teams such as Steakhouse, Gauntlet, K3 Capital, and others deploy MetaMorpho vaults and act as strategy curators, defining each vault’s mandate (which Blue markets are allowed, per‑market caps, concentration and LLTV limits) and then manage allocations and rebalances across those markets over time. They don’t change the underlying market parameters, instead building portfolio‑level risk/return profiles on top of fixed, immutable markets. In simple terms, Morpho commoditizes lending infrastructure; curators differentiate on judgment.

The split is positive-sum when it works. Borrowers get access to liquidity for collateral types or credit profiles that a single global risk committee might never approve. Depositors get delegated risk management instead of having to underwrite every isolated market themselves. Curators can monetize underwriting skills without building, auditing, and bootstrapping a lending protocol from scratch.
Curators earn fees for turning protocol infrastructure into managed credit products, usually taken from the vault's gross yield before it reaches depositors. It is common to also charge a performance fee of ~5-10% on the native yield the vault generates. These fees accrue as the vault's share price rises and are claimed by the curator's designated fee recipient address. In Morpho V2, management fees are capped at 5% annually and performance fees on vault interest are capped at 50%, though institutional curator fees usually sit well below those ceilings.

Across the curator category, total annualized revenue was roughly $12.3M in 2025 across about $2.3B in average daily assets under curation. This figure is still small compared to traditional asset management.

Depositors in a vault allocated to sUSDe and stETH markets might also receive base lending interest from borrowers, forwarded MORPHO token rewards distributed by the DAO, Ethena shard points on USDe markets, and Lido incentives on stETH markets simultaneously. The curator receives none of the token rewards or points. Those flow directly to depositors via Merkl, Morpho's reward distribution infrastructure.
DAO‑run curators play the same portfolio manager role for their own ecosystems with different economics. Protocols like Spark or Sky operate savings‑style vaults that hold native stablecoins and related markets, using vaults to implement mandates set through governance. Governance dictates which venues are allowed, how concentrated the vault can be in their own token, and how aggressively it can loop or rehypothecate. In many DAO‑run vaults, management and performance fees are set to zero or minimal. The protocol indirectly benefits from deeper demand and stickier float for its own assets.

Distribution Layer
Distribution turns vaults into user-facing financial products. It takes an onchain strategy and embeds it directly into channels that already hold deposits: exchanges, wallets, fintech apps, custodians, and treasury platforms. Those frontends hide the complexity and friction of DeFi, exposing a simple unsophisticated action to users who never see a contract address.

Products like Kraken’s DeFi Earn live fully inside the Kraken app, but route deposits into Veda vaults that supply liquidity to Aave and other protocols, with firms like Chaos Labs or Sentora managing the onchain strategy while Kraken owns UX, eligibility, and disclosures. Coinbase’s USDC lending works the same way, pushing in‑app deposits into Morpho vaults curated by Steakhouse while Coinbase handles accounts and regulation.

As an example, Whop Treasury shows that embedded DeFi distribution is already live. Whop users can opt into yield from the same balance screen they already use to receive and manage payouts. Under the hood, eligible balances are converted into USDT, routed through a Veda vault on Plasma, and supplied into Aave’s Plasma market. The end user does not need to choose a protocol, manage a wallet, pay gas, or diligence a vault at the point of deposit. The distributor owns the customer relationship, while the vault provider and lending protocol become invisible infrastructure.
Web3 wallets such as Binance Web3 Wallet, OKX Wallet, and Bybit Wallet follow a similar pattern in a more crypto-native setting. The user sees a one-tap “Earn” or “Lend” button, while the wallet routes deposits into underlying venues such as Aave, Venus, Yearn, or other vault infrastructure. The product label is simple, but the risk still depends on the route underneath.
Distribution is a strong moat. Winning a major exchange or custody mandate brings in more and stickier AUM than winning thousands of small direct depositors.
Failure Modes
Depositors comparing curator vaults usually start with the headline APY figures. A stablecoin vault yielding 8% looks a lot better than another at 5% in the absence of any other information. Yield, however, mostly tells you what the curator is advertising, not how they got there, which collateral is producing that spread, or what the depositor is absorbing if that collateral fails.
In a market where allocators chase headline APY and curators earn on AUM and performance, the cheapest way to print a bigger number is to accept riskier collateral at more aggressive LTVs, not to actually “curate risk” despite what their name suggests.
Stream Finance's November 2025 collapse shows what that pressure produces at its worst. MEV Capital and Re7 both allocated to Stream's xUSD despite a Yearn developer posting a specific public warning months before the collapse. The curators essentially privatize the gains through fees while socializing the losses among depositors, walking away with 6 months of profit before their AUM collapses.

This is the central failure mode for curator vaults. Some information was technically visible or publicly discussed, but the average depositor did not have the equivalent of a fund report disclosing the mandate, current exposures, concentration limits, counterparty dependencies, offchain manager risk, redemption assumptions, and a history of allocation changes. Without that context, depositors were effectively underwriting a curator’s brand and an APY number rather than the actual risk beneath the vault.
Impaired Collateral
Protocols can work exactly as designed and still result in losses for curator vaults if collateral gets compromised.
In March 2026, Resolv’s USR was depegged after a compromised mint key allowed an attacker to mint 80M of unbacked USR to swap for ETH, extracting $25M altogether. The depeg cascaded into every Morpho market that had admitted USR as collateral, producing roughly $180M in cascaded losses across the system.

Gauntlet, Re7 Labs, KPK, and 9summits had all enabled automatic supply as a Public Allocator to Resolv-related markets, rotating capital toward USR markets for hours after the exploit because the onchain utilization signals it read were accurate. The problem, however, was the issuer’s infrastructure, not the market mechanics. No amount of clever rebalancing logic fixed upstream mint and custody design.
Smart Contract Vulnerabilities
Collateral failure produces losses bound by the supply cap a curator set for that market but a protocol contract exploit can drain a vault entirely because the attack vector is the contract logic rather than the value of an underlying asset.
Balancer V2 was exploited for $128M in November 2025 due to a rounding/precision issue in its Composable Stable Pool invariant across multiple networks, and Yearn’s yETH vault lost around $9M to a bug in a StableSwap pool for liquid staking tokens.
Both incidents were accounting logic failures despite being audited systems, illustrating that “number of audits” is a weak proxy for risk. From an allocator’s standpoint, protocol risk is mostly a function of code scope and mutability where small, immutable contracts with narrow functionality are likely safer than sprawling, upgradeable systems where governance or multisigs can push changes to core behavior.
Shared Infrastructure Breaks
In April 2026, a compromise involving KelpDAO’s rsETH and their LayerZero OFT infrastructure triggered simultaneous defensive pauses across 30+ protocols within hours, affecting either their protocol or native token. The list included Ethena, Ether.fi, Curve, WBTC, Kamino, Lombard, Frax, and many more. None had been directly exploited. LayerZero’s DVN, the crosschain messaging layer all of them used for token bridging, was the compromised vector, and every protocol relying on it had to treat its bridged assets as potentially affected until the attack surface was contained.

The failure propagated through a shared dependency that each individual protocol had accepted as external infrastructure. The defensive response was simultaneous across thirty protocols, creating market-wide liquidity and bridging disruption regardless of whether any specific protocol's own code was touched. A depositor in a Kamino vault on Solana that morning was exposed not to a Kamino failure but to an infrastructure failure three layers below what any standard vault risk assessment would have examined.
Advancement of AI in Discovering Exploits
AI changes the exploit surface because the same tools now improve both defense and attack. Security firms like Sherlock, Cantina, and Certik are already using LLM agents to scan code, explain edge cases, and triage findings faster. Attackers can use similar systems to search deployed contracts, bridge code, oracle logic, ZK circuits, and dependency trees at lower cost. The direction of travel is clear: vulnerability discovery is becoming cheaper, faster, and less dependent on a small pool of elite human auditors.
This does not mean every recent exploit was AI-driven. Many large DeFi failures still come from simpler weaknesses: compromised keys, admin permissions, misconfigured multisigs, bridge dependencies, oracle assumptions, and operational mistakes. The point is that allocators should not take comfort in old code merely because it has not broken yet. If AI expands the search capacity of both white-hat and black-hat actors, then dormant bugs that survived years of manual review can become discoverable in days.

The recent Zcash Orchard vulnerability was a warning shot. A security researcher using Claude Opus 4.8 helped identify a critical counterfeiting bug in Zcash’s Orchard shielded pool, a flaw that had existed for roughly 4 years. The issue could have allowed unlimited, undetectable counterfeit ZEC inside the shielded pool and was remediated through an emergency network upgrade.
For onchain asset management, the second-order risk is capital withdrawal. The fear is not only vaults getting exploited, but also allocators reassessing whether the risk premium is high enough for any onchain deployment. The most likely capital reaction differs by user profile:

The practical defense is to reduce the number of things AI can break and limiting how far a break can propagate. Immutable, narrow-scope contracts with battle-tested logic should carry lower risk than complex, upgradeable systems with broad permissions. Isolated markets are safer than pooled contagion when collateral fails. Hard supply caps, withdrawal liquidity buffers, timelocks, circuit breakers, oracle diversity, and explicit dependency maps are important considerations as the cost of finding vulnerabilities falls.
This also means curators need a higher reporting standard. Allocators should demand live exposure reporting, contract dependency lists, admin-key disclosures, bridge and oracle dependencies, concentration limits, max loss assumptions, and a history of allocation changes.
The conditions that would produce AI-assisted exploits at scale are already in place:
- Frontier models like Claude Mythos are highly skilled at cyber-security and hacking tasks, outperforming humans.
- Low marginal cost of AI usage in identifying vulnerabilities against a large payoff if successful.
For allocators this is not a reason to avoid lending to vaults or even DeFi as a broader category. Again, preferring immutable contracts that have been battle-tested over the years has a much lower probability of failure than a freshly deployed, complex protocol.
Catalysts
Agentic-driven Vault Strategies
Curator operations are moving from human‑driven to agent‑driven as they can monitor more markets and react faster. Morpho Agents beta, launched April 2026, introducing User Agents that read, simulate, and write positions across Ethereum and Base, and Builder Agents (AGENTS.md file + skills) that package contract addresses and decision trees so a coding agent can go from instruction to production code in minutes. Over 130,000 AI agents had registered onchain identities since January 2026.
For curators, this accelerates execution but does not affect the strategic layer where the decisions live. Collateral selection, supply cap discipline, and counterparty evaluation are still human judgments. What automation changes is how fast those judgments get executed and how consistently they get enforced. For depositors, the effect is that vaults built on agent infrastructure will respond to market conditions faster, which allows for quick risk-mitigation measures during risk events.
TradFi Participation
Products like BlackRock’s BUIDL, Franklin’s BENJI, and many others, take existing regulated money‑market or Treasury portfolios and wrap them as onchain tokens held by KYCed investors. The underlying allocation and yield are still pure TradFi, but once those tokens can sit in isolated lending markets or allocation vaults, they become yield‑bearing collateral that onchain credit strategies can lend against rather than static assets. Today, tokenized funds enter DeFi by being used as collateral in curated, KYC‑compatible vaults without directly depositing into public pools.
Separately, there is another group of TradFi managers themselves deploying capital into curator strategies. Bitwise is the first major traditional asset manager to launch a curator vault on Morpho in January 2026 targeting institutional‑grade stablecoin yield with its own portfolio team defining risk. Custody platforms like Anchorage and Taurus have integrated Morpho vaults so banks and asset managers can allocate balance‑sheet assets into non‑custodial onchain strategies from within existing governance and compliance workflows. These structures give TradFi LPs an increasingly legally legible way to treat onchain asset management as a fixed-income strategy.
Conclusion
Onchain asset management has evolved from simple yield aggregation into a new intermediation layer for DeFi. The product users see is still a vault, but they function more like portfolio management: selecting collateral, sizing exposure, managing liquidity, and deciding which risks are worth being paid for. DeFi’s opportunity set today is too broad and operationally complex for most users to underwrite.
Isolated markets, ERC-4626 vaults, curated mandates, and embedded distribution have created a structure that can absorb far more capital than the first generation of APY routers ever could. But that’s also the constraint where the category is only as strong as the curator’s underwriting discipline and the infrastructure dependencies beneath it. A higher APY is not evidence of better asset management. In many cases, it is simply evidence that more risk has been accepted somewhere deeper in the stack.
For allocators, the right question is not which vault pays the most today, but which manager has the clearest mandate, the best risk controls, the strongest reporting, and access to sticky distribution.
References
Anthropic — Claude Mythos Preview
Balancer — Incident Update Post
Bitwise — Bitwise Expands Onchain
Chaos Labs — Kraken DeFi Earn Risk Post
CryptoRank — Re7 Labs Criticism
DefiLlama — Onchain Capital Allocators
Ethereum — ERC-4626 Tokenized Vault Standard
Ethereum Foundation — Morpho Deposit Announcement
Fidelity Digital Assets — 2024 Institutional Investors Digital Assets Survey
Gauntlet — Introducing Restaking Vaults on Symbiotic
JPMorgan — JPMorgan Asset Management
Karpatkey — Treasuries Under Management
LayerZero — Incident Update Post
MEV Capital — Stream Finance Post
Morpho — Introducing MetaMorpho
Morpho — Morpho Blue
Morpho — Public Allocator Concepts
Morpho — Understanding Morpho Vaults
Nathan McCauley — Anchorage Morpho Post
Resolv — Postmortem March 22 2026 Incident
Sentora — Kraken Launches DeFi Earn


