Pendle has become the only surviving protocol in the DeFi fixed-rate sector. Tokenized U.S. Treasuries are bringing real yield on-chain, fundamentally transforming the underlying logic of lending and trading.
Written by: Vaidik Mandloi
Compiled by: Chopper, Foresight News
Between 2020 and 2022, a group of sharp-eyed crypto investors bet on a batch of protocols that built on-chain fixed-rate lending products, allowing users to lock in yields similar to government bonds. Element.fi raised $32 million from a16z crypto and Polychain Capital, Notional.finance raised $10 million in a round led by Pantera Capital, and at least five other similar projects followed.
Today, every one of these projects has died. The yields they tried to package into their products were entirely fabricated, mainly because DeFi in 2021 relied on token rewards minted out of thin air by protocols and leveraged looping to manufacture yield. Once users exited en masse, the whole system collapsed.
In this round of industry shakeout, the only surviving protocol is Pendle.finance. The reason it survived is that Treasury assets on RWA.xyz began moving on-chain, bringing real cash flows, which allowed developers to build products around real yield. I believe the same shift is about to hit all of our existing DeFi primitives.
Various Workarounds
To understand this transformation, we first need to understand the market environment in which DeFi was born.
Looking back at on-chain financial markets in 2018 and 2019, the tradable assets were all crypto-native tokens, with no cash flows, no legal entity behind them, and violent price swings that could wipe out 90% of value in a single afternoon.
The participants were anonymous wallet addresses, with no credit history, no legal identity, and no recourse mechanism in the event of default. Ethereum could process only 7–8 transactions per second at the time, with block times of 13–15 seconds. The capital participants were mainly retail speculators, plus a handful of crypto-native funds willing to take extreme risk for high returns.
The on-chain market at the time lacked every basic condition needed for a properly functioning financial market. A mature market needs market makers willing to quote both sides and continuously provide prices, and underlying assets whose price behavior is predictable enough to hedge risk; it also needs lenders able to assess a borrower's ability to repay, which requires knowing the borrower's real identity.
DeFi back then was far from meeting that standard. Assets were too volatile, professional market makers were unwilling to participate at all, borrowers were anonymous wallets, and the user base had no interest in financial products with fixed maturities.
So developers at the time began designing various stopgap solutions. For example, Uniswap invented the automated market maker (AMM) not because it was a better way to trade. The AMM was born because no one was willing to make markets in tokens that could crash 90% within minutes. The solution was to write pricing logic directly into smart contracts, let anyone inject capital into a pool, and have the constant product formula automatically handle price discovery, with every trade automatically adjusting the exchange price based on the ratio of tokens in the pool.
Many DeFi-native users still do not fully understand that this mechanism is fundamentally different from traditional market making. Traditional market makers earn the bid-ask spread. But AMM liquidity providers (LPs) are doing something entirely different. Whenever arbitrageurs spot a price gap between the pool and the external market and trade to close it, the LP on the other side is effectively paying the arbitrageur to rebalance its own portfolio.
For example: you deposit equal values of ETH and USDC into a Uniswap pool. Suppose ETH rises 20% on Binance, but the pool has not yet registered that price change. An arbitrageur will spot the cheap ETH in the pool, buy it, and transfer it to Binance to sell for a profit. In the pool, USDC increases and ETH decreases, which amounts to you selling ETH on the way up.
When ETH falls, the reverse happens: arbitrageurs dump ETH into the pool and drain USDC. Trade after trade, your portfolio automatically reverts to a 50/50 allocation, and arbitrageurs take the difference as profit. The math behind this is quite interesting: even under the extreme assumption that every trade in the pool is pure arbitrage with no retail trading at all, liquidity providers can still earn positive returns after the small fees charged on each trade.
Aave and Compound followed the same constrained design logic in lending. Ordinary banks typically assess your ability to repay using credit scores and income records. But that is impossible with anonymous on-chain wallets. So these protocols adopted overcollateralized lending.
Perpetual contracts replacing traditional dated futures came about for the same reason. Rolling quarterly futures contracts requires counterparties to perform on fixed settlement dates. But in crypto markets, where anonymous capital rotates across protocols chasing the highest yield, such long-term performance commitments simply cannot be guaranteed.
Of course, these workarounds did produce usable products. But zoom out and look at the financial system they collectively built, and the problem becomes glaring: the collateral underpinning all these products generates no yield itself.
We can use Hyman Minsky's theory to stress-test any financial arrangement: examine whether the income generated by the borrower's collateral assets is sufficient to cover the debt. Financial relationships can be divided into three categories.
The first is hedge financing, the safest: asset income covers both interest and principal. For example, a salaried worker repaying a mortgage with wages is a typical case.
The second is speculative financing, an intermediate state: income can cover interest, but when principal comes due, refinancing is needed to roll it over. Companies rolling over corporate bonds fall into this category.
At the bottom is Ponzi financing: income covers neither interest nor principal. The only hope of staying solvent is that the price of the assets held keeps rising; once prices stop rising, the whole system collapses.
Now look at a user borrowing on Aave: the borrower deposits $10,000 worth of ETH, borrows $6,000 in USDC, and uses the stablecoins freely. But the ETH in Aave's vault generates no income to repay the loan. The only thing keeping the loan safe is that ETH's market price stays above the liquidation line. If the drop is large enough, the protocol liquidates the collateral and closes the position.
The entire lending arrangement depends entirely on asset prices holding up until the borrower exits voluntarily. Almost all DeFi overcollateralized lending using crypto-native tokens as collateral falls into this category.
How the Transformation Will Happen
Let's replace the collateral in the scenario with tokenized U.S. Treasuries. The lending protocol and the borrower stay the same, but the collateral is no longer ETH; it is $10,000 worth of tokenized Treasuries yielding 4.5% annually, and the borrower takes out stablecoins at a cost of 3%. The collateral generates $450 a year, while the loan costs only $300. The collateral's cash flow covers the cost of the debt. This yield is unrelated to crypto market cycles; it comes from U.S. government bond coupons, and the cash flow keeps flowing whether Bitcoin is at $100,000 or $30,000. Every additional day the loan position remains open, the borrower's position becomes more stable. This stands in stark contrast to ETH-collateralized borrowing: with every market decline, the borrower moves one step closer to liquidation, and the collateral itself generates no cash flow.
When the vast majority of collateral in DeFi yields nothing and depends entirely on price appreciation, all positions are exposed to the same one-directional risk. One asset class falls, triggering liquidations; the selling pressure from forced liquidations pushes prices down further, triggering more liquidations and creating a cascading stampede. We have seen this process in almost every crypto downturn since 2020. People always treat it as a black swan event, but when collateral itself generates no cash flow, this outcome is entirely expected. But when a substantial portion of collateral consists of yield-bearing assets like U.S. Treasuries, positions remain stable even if the crypto market crashes, because their cash flow is independent of the crypto market.
For the first time, the lending system has a safety floor that does not depend on a bull market. The scale of tokenized money market fund shares grew from $770 million at the end of 2023 to $14.82 billion on October 5, 2026, an increase of about 19x in less than three years.
There is precedent for this model. In 1955, Midland Bank in London began accepting dollar deposits. At the time, the U.S. Regulation Q capped interest rates on U.S. bank deposits, while British banks were not subject to that constraint. Dollar funds flowed overseas in pursuit of higher yields, and a parallel dollar system was born, reaching about $4.7 trillion by the early 1980s.But the Fed did not suppress the Eurodollar market; it chose to accommodate it. A parallel market born from the limitations of the existing system will continue to expand until the original restrictions are removed. Thus Regulation Q was gradually repealed, money market funds emerged as a domestic American competitor, and this parallel dollar system was ultimately absorbed into the mainstream system.
Stablecoins are this generation's Eurodollars, and tokenized Treasuries are this generation's money market funds.
Those defunct protocols tried to build interest rate markets on-chain before real interest rates existed, using token issuance and leverage loops as a substitute for real yield. But you cannot build a complete yield curve on yield that can be shut off at any time by a protocol governance vote.
Pendle survived because it waited until tokenized Treasuries and yield-bearing stablecoins emerged, when the market finally had government-backed real cash flows on which to build products. Pendle's Boros is now the first on-chain interest rate swap product that actually works.
The same logic applies to the trading track. In liquid trading pairs, AMM liquidity providers lose about 11% of their capital per year to arbitrage; for tokens not listed on centralized exchanges, bearing this "arbitrage tax" was once the only option. But for assets like Treasuries, equities, and bonds that already have extremely deep global liquidity, taking the AMM route is highly irrational.
Such assets require on-chain order books with performance comparable to centralized exchanges. For years, major public chain teams claimed this was technically impossible, but Hyperliquid disproved that conclusion. In early 2025, it at one point handled roughly 60% of on-chain perpetual contract trading volume.
All these signs indicate that the infrastructure built for tokenized Treasuries and tokenized equities will hew closer to traditional electronic trading markets than many people might expect, given that these assets were originally the service targets of traditional financial markets.
The first generation of DeFi primitives was born under the conditions of anonymous wallets holding highly volatile, cash-flow-free tokens. But now the underlying assets have changed, and these past stopgap measures will gradually exit. The new infrastructure that replaces them will converge with the traditional market architecture that has processed trillions of dollars in daily trading volume for decades.