Everyone keeps asking the SEC what a vault is. More useful question: what should a vault be, now that the constraints that produced the shared-vault container no longer exist?
Billions sit in these things. Banks and fintechs are shopping. Regulators are circling. The industry is arguing loudly, and correctly, that the code matters more than the label. True. And while that fight has absorbed most of the oxygen, the product itself has barely moved. The architecture most retail users touch today was built for problems that no longer exist.
What is good
The real wins are architectural. Collateral rules enforce themselves. Liquidations run without a phone call. Interest comes out of a utilization curve. Settlement clears in a block. The same rails flatten access: a balance in a neobank app and one on a corporate treasury desk hit the same market under the same terms. And in a well-written vault, “trust me” gets replaced with “you cannot.” A smart contract that was never given the function to send funds to an outside address literally cannot. The protection shifts from behavioral to physical.
What is bad
“Vault” is a wrapper, not a product. The same contract shape can sit on top of a morpho-style supply or a leveraged basis trade routed through five venues with an operator who can pause your withdrawal. The screenshot looks identical, and depositors cannot tell them apart.
What gets sold as composability also builds leverage no one is tracking. A vault share becomes collateral somewhere else, which becomes collateral somewhere else again, and a bad print on page one shows up as a liquidation on page three in a protocol that has never heard of page one. The 2008 word for this stack was CDO. We just removed the gatekeepers who had to sign off before you shipped one.
Onchain and verifiable are not the same word. Reach into private credit, invoices, or a CEX desk, and the chain is just reading back whatever number the operator or oracle hands it. The depositor sees a clean dashboard. Sometimes the number is the operator grading its own test.
What is ugly
The ugly starts when the architecture is used to launder discretion as self-custody. Material decisions quietly leave the chain: the surface is code, but the marks, risk gates, and allocation calls live in someone’s spreadsheet. Capital gets steered by economics the depositor cannot see, as curators, platforms, and issuers trade rebates and preferred allocations for flow.
Worse, in some versions of vaults, the curator or infrastructure provider takes fees from the liquid sleeve while the illiquid sleeve sits on an impairment the depositor cannot easily observe. The performing collateral pays the manager; the depositor eats the bad collateral.
Why pooling stopped making sense
I wrote the fuller version of this in Crowdfund Insider. Short version: pooling was a workaround. Gas was expensive, UX was brutal, and the only way to get retail into yield was to bundle their capital and let a strategist drive. The constraints are gone. L2 gas is a rounding error. Bridges move stablecoins in seconds. MPC and account abstraction let a single user hold and rebalance a diversified book themselves. The container outlived the problem.
The exit problem is worse than the yield problem. Lose confidence in one lending market, one RWA issuer, one looped position, and you cannot leave it in some vault structures. You can only leave the whole vault, at the infrastructure providers/curators’ mark on the illiquid queue, in the vaults queue. You picked a yield number and were told you had custody.
Single-strategy vaults are more efficient, and it is not close
Pooled vaults do two different jobs. Capital routing is a solved engineering problem. Portfolio construction is a judgment business and should belong to the depositor or someone the depositor explicitly hired. A single-strategy vault commits to one job in plain language: one destination, one asset, one pricing model, one exit. The dollar in is the dollar earning yield. The APY on the page is the APY in the wallet, less the shown fee. If confidence wanes, the depositor closes that one bet; everything else stays untouched.
A multi-strategy vault blends every position into one share price and gives the depositor one lever: in or out. In exchange they absorb undeployed-balance drag, cross-subsidy between strategies, fees skimmed from whichever sleeve is still spendable, and an exit queue set by the slowest asset in the pool. The container is running a portfolio the depositor never signed up for and cannot take apart.
Where this goes
Every depositor gets their own wallet. Capital moves into named, single-purpose destinations they explicitly chose. The deposit starts working on arrival. Fees are quoted inside the destination product, as a fee. You can shut off any position without disturbing the rest of the book. Nobody drives the car on the depositor’s behalf, and nobody stands between the depositor and the exit.
That is the architecture Ground is building: every depositor in their own MPC wallet, wallet-agnostic rails, and a shelf of named strategies they can turn on, swap, or shut down one at a time. The platform’s job is routing, custody, and execution. The platform’s non-job is deciding which bet to be in.
Vaults are neither the enemy nor the hero. They are a container that was useful for one phase of this market and got pressed into jobs it was never designed for. Shrink them back to what they are good at, hand everything else to real products and real accountability, and stop pretending the container is the breakthrough. The breakthrough was always the depositor having direct, deployed, single-click control over where their yield comes from.
Stephanie Vaughan is the COO of Ground, a company building the Stripe-like interface for TradFi institutions (neobanks, Fintechs, and wealth advisors) to access on-chain assets like DeFi and real-world assets (RWAs) without ever touching blockchain infrastructure themselves. Ground’s platform uses non-custodial MPC wallets so partners can offer their end users tokenized yield products, RWAs, and stablecoin-native services while maintaining full custody and compliance. Previously, Stephanie co-founded and served as COO of Veda, the DeFi engine that has secured over $6 billion in digital assets for leading protocols and Fintech platforms. Earlier, she held C-suite roles across Web3, including xx Network (founded by cryptography pioneer David Chaum), and was in investment banking at Houlihan Lokey. Stephanie is a U.S. Naval Academy graduate, a Columbia MBA, and a former Marine Corps Captain. You can connect with Stephanie on X.