DeFi Lending Has a $53 Billion Efficiency Problem

There's $53 billion locked in DeFi lending protocols right now. That number sounds impressive until you look at how much of it is actually doing useful work.

A 1inch report from late 2025 found that between 83% and 95% of liquidity sitting in major DeFi pools is idle. Not earning fees, or generating any type of returns...just sitting there. Half of all liquidity providers are losing money after accounting for impermanent loss, with net deficits above $60 million. One Uniswap v3 pool alone saw over $30 million in lost profits from just-in-time liquidity manipulation.

And this is happening while lending protocols collectively pulled in ~$20 million in fees last week, while stablecoin supply hit record highs above $310 billion. The infrastructure has never been bigger. The capital has never been less efficiently used.

I think DeFi's capital efficiency problem gets talked about in the wrong terms. Most people frame it as a user experience issue, or a gas optimization problem, or something that'll get fixed with better interfaces. It's actually an architecture problem. And it runs deeper than most of the current solutions are willing to admit.

The liquidation tax nobody accounts for

Here's where the inefficiency gets personal.

In most lending protocols, risk management works like a light switch. Your position is either safe or it's getting liquidated. There's no dimmer. An academic study of Aave, Compound, MakerDAO, and dYdX — protocols that collectively handle over 85% of Ethereum's lending market — found that their liquidation mechanisms are designed to protect lenders first, borrowers second. That's rational. But the side effect is that liquidations sell off excessive amounts of discounted collateral at the borrower's expense.

During the March 2020 crash, MakerDAO liquidators pulled $13 million in profit in a single month because their bots couldn't keep up with the chaos and human liquidators swept in with lowball bids. In October 2025's "10/10" flash crash, $3 billion in positions were liquidated almost instantly. Those losses didn't just hurt the individual borrowers. They cascaded through the system, triggering more liquidations, which triggered more selling, which triggered more liquidations.

The binary nature of these systems means borrowers can't partially de-risk. They can't adjust gradually. They ride the position until the protocol decides they're done, and the protocol's decision is all-or-nothing. For anyone running a leveraged position through a volatile week, this isn't a theoretical flaw. It's the thing that wipes you out 2% past your liquidation threshold when a graduated adjustment would have kept you in the game.

Collateral is still stuck in 2020

The other half of the problem is what lending markets will actually accept.

Most protocols support the same narrow set of assets: ETH, BTC, major stablecoins, maybe a handful of governance tokens. This made sense in 2020 when DeFi was small and every collateral type was an existential risk. It makes less sense now.

Liquid staking tokens, yield-bearing stablecoins, tokenized treasuries, structured vault positions — these are real assets with real value, and in many cases they're already generating yield on their own. But the majority of lending protocols force you to choose. You can either earn yield on your staked ETH, or you can use it as collateral to borrow against. Not both. So capital that could be working in two places at once gets stuck in one.

The yield-bearing stablecoin market alone has doubled in supply over the past year. Tokenized real-world assets crossed $18 billion. BlackRock's BUIDL fund, Franklin Templeton's on-chain money market fund, Ethena's USDe — these aren't experimental anymore. They're institutional-grade instruments sitting at the edge of the lending ecosystem, waiting for protocols that know what to do with them.

The protocols that figure out how to accept productive capital as collateral, without blowing up their risk models in the process, are going to unlock a category of efficiency that the current generation can't touch.

What "better" actually looks like

The next wave of DeFi lending isn't solely about incremental improvements to existing designs, as the core architecture needs to change in a few specific ways.

Risk management should behave like a curve, not a cliff. Positions should be able to adjust gradually as conditions shift, with partial de-risking instead of full liquidation. This alone would reduce the cascading sell-offs that amplify every market downturn and eat into borrower equity unnecessarily.

Collateral should be allowed to keep working. If you deposit a yield-bearing asset, the yield shouldn't disappear into a void. A well-designed system would recognize that productive collateral is, by definition, better collateral — it's growing in value while it secures your loan.

And the system itself should be composable enough to handle new asset types as they emerge. The pace of innovation in tokenized assets, restaking derivatives, and structured products is accelerating. A lending protocol that only works with the assets people used two years ago is already falling behind.

These aren't incremental upgrades. They require rethinking how capital, risk, and collateral interact at the protocol level.

Where this is going

The DeFi lending market in 2026 looks very different from where it was even 18 months ago. Aave is pushing toward V4 with a unified liquidity layer. Morpho is experimenting with modular vaults and peer-to-peer matching. Pendle has opened up structured yield markets that didn't exist before. The entire sector is converging on the same realization: the current architecture has a ceiling, and we're hitting it.

The protocols that break through that ceiling will be the ones that treat capital efficiency as a design principle, not a feature. The ones that build risk systems with enough nuance to handle real market conditions. The ones that let productive assets stay productive, even when they're being used as collateral.

That’s the gap Curvance is built to close.

Not by iterating on the same models, but by rethinking how capital behaves inside a lending system from the ground up. Instead of forcing users to choose between deploying capital and earning on it, Curvance is designed around the idea that productive capital should remain productive at all times, and yield-bearing assets become first-class collateral.

At the same time, risk management isn’t treated as a binary switch. Curvance introduces a more nuanced system where positions can be adjusted progressively, reducing the need for abrupt, value-destructive liquidations. The goal is to protect lenders, as well as preserve borrower capital and stabilize the system as a whole.

And then there’s the piece most protocols still ignore: where the value actually goes. Traditional liquidation systems leak the majority of incentives to external actors through MEV. Curvance flips that dynamic by an auction-based liquidation system that aim to internalize that value back into the protocol, improving overall capital efficiency rather than extracting from it.

The result is a lending layer that not only grows TVL, but it scales utilization, preserves value, and compounds efficiency at the system level.

Because the next phase of DeFi isn’t about more capital.

It’s about making the capital already on-chain actually work.