Three Firms Control 56.9% of Curated Vault TVL. Only 5.7% Touches the Real World.

There is a vault holding $281 million of stablecoins on a blockchain owned by Robinhood. Robinhood did not build the strategy. A firm called Steakhouse Financial did: a crypto-native asset manager most people have never heard of, now running $2.5 billion across 67 vaults on six chains. Kraken’s chain, Ink, holds another $688 million in just five vaults, all run by one firm. Coinbase’s Base holds $1.9 billion more. Roughly 30% of professionally curated vault capital now sits on chains owned by regulated consumer platforms.

That is the onchain vault industry in August 2026, and almost nobody outside of it is watching.

Everything below comes from a vaults.fyi export pulled August 7, 2026: 1,231 vaults holding $78.2 billion, of which $9.9 billion sits in 577 vaults run by 112 named curators.¹ Three conclusions up front:

• Three firms control 56.9% of curated vault TVL. The median curator runs $8.2 million.
• Curated stablecoin vaults out-yield passive lending by roughly 114 basis points, and about 97% of that yield is intrinsic strategy return, not token emissions.
• Roughly 94% of curated capital sits in crypto-native strategies. RWA-tagged vaults hold just 5.7% of the money.

What a vault actually is

A vault is a smart contract that automatically invests your deposit according to allocations someone else sets based on their investment experience. It is more or less a new way of investing with an asset manager: you deposit, the strategy runs, you own a proportional share of everything inside, and you can withdraw on the vault’s own terms. Think money market funds, ETFs, separately managed accounts: the same job, rebuilt as software that settles onchain.

The person setting those allocations is called a curator. Curators do what portfolio managers do: pick markets, set risk limits, rebalance, and charge fees for it (among the 226 curated vaults that disclose fees, the median performance fee is 10%). The difference is the wrapper. Mutual funds packaged active management for the mail-order era. ETFs packaged it for the exchange era. Vaults package it for programmable settlement: typically instant issuance and redemption, transparent holdings, and distribution through any app that can talk to a blockchain.

The wrapper is growing fast. Vault TVL ran from $2.5 billion in early 2025 to more than $12 billion across trackers in 2026, and Kiln projects $500 billion in vaults by 2030.² Inside this dataset alone, 173 surviving curated vaults launched in the first seven months of 2026, already carrying $3.3 billion: a third of all curated TVL deployed this year.

Three firms, then everyone else

The curator league table is a barbell. At one end, three firms: Steakhouse Financial ($2.48B), Sentora ($1.77B), and Gauntlet ($1.39B), together 56.9% of all curated TVL. At the other end, a long tail: 60 of the 112 active curators run under $10 million, and only 16 exceed $100 million.

The Herfindahl index on curated TVL is 1,255: moderately concentrated by DOJ convention, established enough to have winners, open enough that new entrants keep arriving. One caveat on counts: the export applies a minimum-TVL filter, so 112 curators and 577 vaults are floors, not totals. The TVL figures capture essentially all meaningful value; the vault counts undercount the dust.¹

The top three also show two different playbooks. Steakhouse and Gauntlet run wide: 67 and 64 vaults across many chains. Sentora runs deep: 18 vaults averaging $98 million each, three to four times the average vault size of its larger rivals. Both models cleared $1 billion. There is no single shape a winning curator has to take.

Half the plumbing runs through one protocol

Curators build on vault infrastructure protocols, and one of them dominates. Morpho carries $4.93 billion, 49.8% of all curated TVL, across 248 vaults. Veda is a distant second at $1.32 billion, Euler third at $0.67 billion. And within Morpho, Steakhouse alone curates $2.46 billion: exactly half of Morpho’s curated book.

So the market’s nesting doll looks like this: half the industry runs on one protocol, and half of that protocol’s curated deposits answer to one firm. A fee change or a technical failure at Morpho is a systemic event for the entire category as currently built.

Geography tells the more interesting story. Ethereum mainnet still holds 53.7% of curated TVL, but the fastest-growing shelf space belongs to exchanges and brokers: Base at $1.92 billion, Ink at $688 million, Robinhood’s chain at $334 million. Curators are becoming the asset-management back end for fintech front ends, the way BlackRock sub-advises products that carry someone else’s brand.

The yield is real, with one asterisk

With the structure covered, the question is yield. On the dominant asset class, curation earns its fee. TVL-weighted 30-day APY on curated stablecoin vaults is 4.60%, against 3.46% for uncurated passive lending on Aave, Sky, and peers: a curation premium of 114 basis points.³

More important than the level is the composition. Of the cleaned 3.64% weighted APY across all curated vaults, 3.52 points come from base strategy yield and just 0.12 points from token incentives. About 97% of curated yield is intrinsic. That is a sharp break from the 2021 model, where returns were mostly emissions dressed up as yield.

The asterisk: TVL is a measure of deposits, and deposits do not always earn. The second-largest curated vault in the dataset, Concrete’s $590 million Delta WeETH vault, prints 0.00% 30-day APY, and nearly $1 billion of top-10 curated TVL currently earns almost nothing. Treat TVL as shelf space, never as value delivered.

The shelf is nearly empty

Here is the number I keep coming back to. Stablecoins are 69.6% of curated TVL, versus 38.7% of the whole export. ETH flips the other way: 51.0% of all vault deposits, just 17.3% of curated. Curators have effectively ceded staking to Lido and taken the managed-cash-yield business for themselves.

But look at what that cash actually earns yield from. RWA-tagged vaults hold $566 million: 5.7% of curated TVL, in 53 vaults. The other 94% earns its return from crypto-native activity: lending to traders, looping collateral, market-making inventory. That matches the broader estimate that about 90% of DeFi vault deposits are crypto-correlated.² When crypto leverage demand falls, so does the yield, at exactly the moment depositors want stability most.

The rails are built. The curators are staffed. The shelf for yield that does not depend on crypto is 94% empty.

Put the pieces together:

• Distribution exists: 30% of curated TVL already sits on Coinbase, Kraken, and Robinhood chains, in front of tens of millions of retail accounts.
• Professional management exists: 112 curators, three of them above $1 billion, charging real fees for real yield.
• The product gap is the asset side: institutional-grade private credit assets, income-producing and largely uncorrelated to crypto, are almost entirely absent from the shelf.

In traditional markets, the money market fund won because a new wrapper met an asset everyone needed and nobody had packaged well. Vaults have the wrapper and the distribution. The non-crypto-correlated asset is the missing piece. O2O Capital Partners manages regulated, DeFi-composable vaults for eligible institutional allocators, holding institutional-grade private credit assets sourced from established institutional partners, which is the side of the trade this data says is empty.

Footnotes

1. vaults.fyi API v2 export, pulled 2026-08-07 14:28 UTC: 1,231 vaults across 21 networks and 78 protocols; $78.16B total TVL, of which $9.90B across 577 vaults carries a named curator. The API applies a default minimum-TVL filter, so vault and curator counts are floors while TVL coverage is essentially complete. TVL is unadjusted for vault-of-vault nesting, so aggregates are upper bounds. Coverage is EVM-centric; Solana protocols such as Kamino are outside this dataset. Vault and curator figures vary materially across trackers by scope: whether a source counts only vaults with a named curator, applies a minimum-TVL filter, reports a single protocol’s dashboard, or nets out vault-of-vault nesting. Headline totals from different sources are not comparable until the definition is matched; the figures in this piece are defined as above. ↩ ↩²

2. Industry growth figures ($2.5B in early 2025 to $12B+ in 2026) per DeFi Llama and public vault trackers; the $500B-by-2030 projection is Kiln’s. The ~90% crypto-correlated deposit estimate is O2O’s own, derived from public tracker data, and is consistent with this export’s 94% non-RWA share of curated TVL. ↩ ↩²

3. APYs are 30-day trailing and net of fees per vaults.fyi methodology. Weighted figures exclude three sub-$1M dust vaults printing annualization artifacts above 100% APY, which would otherwise inflate the curated weighted average from 3.64% to 10.87%. ↩