Tokenized assets need a price-feed safety net before institutions commit DeFi capital

DTCC's tokenization trial now includes roughly 40 firms – JPMorgan, Goldman Sachs, BlackRock, Vanguard and the NYSE among them – testing how to represent shares and Treasuries on-chain. The catch is that those tokens become usable collateral only when a lending market can answer one question: who prices them, and what happens when the venues behind that price go silent.

DefiLlama puts on-chain real-world asset market cap above $51 billion. Those same assets generate only about $3.8 billion in DeFi active total value locked, a utilization rate near 7.7%.

A lending market needs a feed, a set of venues the feed draws from, and rules for what happens when those venues stop pricing tokenized stocks, bonds or gold.

Someone has to choose the oracle, test its independence, cap exposure, and decide when liquidations trigger.

Matthew Fisher, CEO of Katana Network, said an oracle's configuration starts with the venues it pulls price data from at launch. Teams upgrade it as liquidity migrates toward newer or deeper venues. For newly listed tokens, that upgrade lags – liquidity has not concentrated in any single trusted venue yet.

Fisher said institutions delegate that vetting to professional curators, vault operators like Steakhouse and Gauntlet who evaluate collateral and set exposure limits on Morpho, or to protocols like Aave that build their own oracle relationships directly.

“The institutions appreciate that there is a professional kind of in the room,” he said.

A December 2025 study on decentralized credit found a small number of curators managing ERC-4626 vaults now intermediate a disproportionate share of total value locked, concentrating underwriting decisions in that layer of the stack. Fisher's account of institutional behavior lines up with what the data already shows independently.

He said a single oracle manipulation inside one market a curator trusted can taint that curator's entire track record. A curator carrying a damaged record into an investment committee gets what Fisher called “a hard no,” regardless of how it performs elsewhere.

Fisher described the curator as the party that owns the risk decision, absorbing the reputational and commercial fallout when a market breaks. The depositor typically absorbs the financial loss directly. Pool-based models like Aave or isolated markets on Morpho often leave the underlying protocol with no direct liability at all.

April's KelpDAO exploit puts that mismatch on display. Aave governance estimated $230 million in bad debt from the related rsETH position, which originated outside Aave's own codebase. Aave's Umbrella module absorbed about $50 million as a first line of defense.

That accountability gap raises a concern about institutions trusting curators whose primary penalty for a bad call is reputational, while the depositor eats the first dollar of loss. First-loss capital, mandatory insurance, fee clawbacks and auditable exposure disclosures are the kinds of demands that could close it.

Bitcoin trades continuously across deep global venues, so its oracle design centers on aggregation and manipulation resistance. Tokenized equities, bonds and commodities inherit a market calendar their reference asset still observes.

Fisher said there is “not an objective right approach” to pricing those assets once the primary market closes.

Some platforms compute a moving average from market-maker quotes once trading halts. Binance historically leaned on funding rates to influence weekend pricing before announcing new plans for that approach this year. Katana routes gold, silver and oil through Chainlink and closes those markets to new positions once the underlying exchange closes. Traders can still reduce existing positions, and isolated margin contains any losses that follow.

The London Stock Exchange plans a night-time session, LSE 24, for 2027. Nasdaq is moving toward 23-hour weekday trading, and Cboe proposed 23×5 US equity trading. Weekends, trading halts and asset-specific gaps sit outside all three plans.

In the bull case, platforms standardize off-hours pricing, circuit breakers, first-loss capital and curator disclosures over the next several years. Citi projects tokenized assets reaching $8.2 trillion by 2030 under its bull scenario. If DeFi utilization climbs toward 12% to 18% in that world, RWA-linked DeFi active TVL could land near $1 trillion to $1.5 trillion, turning tokenized Treasurys, equities and commodities into genuine collateral primitives.

In the bear case, tokenization keeps expanding in issuance terms without solving its governance layer. Citi's bear scenario puts tokenized assets at $2.7 trillion by 2030. If DeFi utilization stays in the 2% to 4% range that today's data implies, RWA-linked DeFi active TVL lands closer to $54 billion to $108 billion. Tokenized assets pile up on balance sheets, and DeFi lending and composability barely touch them.

Fisher noted that institutional oracle sensitivity for tokenized equities, bonds, or commodities whose underlying markets close on weekends is higher than sensitivity to crypto-native assets.

Institutions need a governance stack around their price feeds durable enough to survive an investment committee, along with a settled answer for who absorbs the loss the day a feed gets it wrong.

RWA returned to active DeFi use within 100 days of the KelpDAO shock, spreading across multiple chains and collateral types.