By Vaidik Mandloi
Translated by Chopper, Foresight News
Between 2020 and 2022, a group of crypto investors backed protocols built around fixed-rate lending on-chain, aiming to offer users something closer to bond-like returns. 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 projects followed.
According to the article, every one of those projects eventually died. The problem was not packaging. The yield inside those products was not real. In 2021, DeFi returns largely came from token incentives minted by protocols and from leverage loops. Once users rushed for the exit, that structure broke down.
The piece says the only protocol left standing from that wave is Pendle.finance. The reason, in the author’s account, is that treasury assets tracked on RWA.xyz started moving on-chain, bringing real cash flow with them. That gave builders something durable to structure products around. Mandloi’s broader argument is that the same change is about to hit most of DeFi’s existing base components.
The first DeFi stack was shaped by the limits of early on-chain markets
The article starts with the market conditions of 2018 and 2019. Tradable on-chain assets were mostly crypto-native tokens. They carried no cash flow, had no legal entity behind them, and were extremely volatile. A token could lose 90% of its value in a single afternoon.
Market participants were anonymous wallet addresses with no credit history and no legal identity. If someone defaulted, there was no practical enforcement mechanism. Ethereum processed only 7 to 8 transactions per second at the time, with 13- to 15-second block times. The capital base was mostly retail speculators and a small number of crypto-native funds willing to take very high risk in search of very high returns.
In Mandloi’s telling, the core conditions needed for a mature financial market simply were not there. Functional markets need market makers willing to quote both sides continuously. They also need underlying assets whose behavior is predictable enough to hedge. Lending markets need lenders to know who the borrower is and to judge repayment capacity from that information.
Early DeFi failed those tests. Asset volatility was too high for professional market makers. Borrowers were anonymous wallets. Users had little interest in fixed-maturity financial products. Builders responded by designing around those constraints.
AMMs, overcollateralized lending and perpetuals were workaround designs
The article uses Uniswap as a central example. Automated market makers, it says, were not invented because they were a universally better trading model. They emerged because almost no one wanted to make markets in tokens that could collapse 90% within minutes. The pricing function was moved into a smart contract, anyone could supply liquidity, and a constant-product formula handled price discovery. Each trade changed the token balance in the pool and updated the exchange rate.
Mandloi stresses that this is fundamentally different from traditional market making. A traditional market maker earns the bid-ask spread. An AMM liquidity provider is doing something else. When arbitrageurs spot a gap between the pool and external markets and then trade it away, LPs are effectively paying for the rebalancing of their portfolio, while arbitrageurs take the profit.
He gives a simple example. A user deposits equal values of ETH and USDC into a Uniswap pool. If ETH rises 20% on Binance and the pool has not adjusted yet, an arbitrageur buys the underpriced ETH from the pool and sells it on Binance. Inside the pool, USDC goes up and ETH goes down. The LP has, in effect, sold ETH on the way up.
If ETH falls, the opposite happens. Arbitrageurs sell ETH into the pool and withdraw USDC. Trade after trade, the LP drifts back toward a 50/50 allocation while the arbitrageur captures the spread. The article notes that, mathematically, LPs can still end up with positive returns after small trading fees even under an extreme assumption that every trade is pure arbitrage and none of the volume comes from retail flow.
The same logic, the piece says, shaped lending protocols such as Aave and Compound. Banks usually underwrite loans with credit scores and income records. Anonymous wallets cannot offer that. The result was overcollateralized lending.
Perpetual futures replacing fixed-expiry futures came from the same set of constraints. Quarterly futures require counterparties to honor obligations on a set settlement date. In crypto, anonymous capital rotates across protocols in search of the highest return, so long-dated commitments are difficult to rely on.
Those designs produced usable products. But viewed as a system, they all sit on the same weak base: the collateral supporting them does not generate income.
A Minsky reading of DeFi collateral
The article applies Hyman Minsky’s framework as a stress test for financial structures, asking whether the income generated by collateral is enough to cover debt obligations.
Hedge finance is the safest form. Asset income covers both interest and principal. The article uses wage income servicing a mortgage as the example.
Speculative finance sits in the middle. Income covers interest, but principal must be refinanced at maturity. Rolling corporate debt fits this category.
Ponzi finance is the weakest. Income covers neither interest nor principal. The structure survives only if the asset price keeps rising. When that stops, the system unravels.
The article maps this framework onto a standard Aave position. A borrower posts $10,000 worth of ETH and borrows $6,000 in USDC. The issue is straightforward: the ETH sitting in the vault produces no income to service the loan. The position remains safe only if ETH stays above the liquidation threshold. If the price drops far enough, the protocol liquidates the collateral and closes the position.
By that logic, the arrangement depends on the asset price holding up until the borrower exits voluntarily. The article argues that nearly all overcollateralized DeFi lending backed by crypto-native tokens falls into that bucket.
What changes when the collateral is tokenized U.S. Treasuries
Mandloi then swaps the collateral. Keep the same borrower and the same protocol, but replace ETH with $10,000 of tokenized U.S. Treasuries yielding 4.5% annually, while the borrower funds in stablecoins at a 3% cost.
That collateral now produces $450 a year, while funding costs are $300. The cash flow from the collateral covers the cost of the debt. Just as important, that cash flow does not depend on the crypto cycle. It comes from U.S. government bond coupons. Whether Bitcoin is at $100,000 or $30,000, the cash flow keeps arriving. The longer the position stays open, the more stable it becomes.
That stands in sharp contrast to ETH-backed borrowing, where each market drawdown pushes the borrower closer to liquidation because the collateral itself generates nothing.
When most DeFi collateral yields nothing and relies on price appreciation, positions are exposed to the same one-way risk. One asset class drops, liquidations start, forced selling pushes prices lower, and more liquidations follow. The article says this pattern has appeared in almost every crypto downturn since 2020. It is often treated as a black swan. In the author’s view, it is exactly what one should expect when collateral has no cash flow.
If a meaningful share of collateral is shifted into yielding assets such as U.S. Treasuries, positions can remain stable even during a crypto crash because their supporting cash flow sits outside the crypto market.
The article argues that this gives lending markets an anchor that does not require a bull market. It cites tokenized money fund shares growing from $770 million at the end of 2023 to $14.82 billion on Oct. 5, 2026, a roughly 19-fold increase in less than three years.
Stablecoins and tokenized Treasuries are framed as a new Eurodollar era
The piece brings in a historical parallel. In 1955, Midland Bank in London began taking U.S. dollar deposits. At the time, Regulation Q capped deposit rates at U.S. banks, while British banks were not subject to that ceiling. Dollar capital moved offshore in search of better yields, and a parallel dollar system emerged. By the early 1980s, it had reached about $4.7 trillion.
The Federal Reserve did not suppress the Eurodollar market. It adapted to it. Mandloi’s point is that when a parallel market is created by limits in the existing system, it tends to keep growing until those limits are removed. Regulation Q was eventually phased out, money market funds emerged in the United States as a competing domestic product, and the parallel dollar system was absorbed into the mainstream.
From there, the article draws a direct analogy: stablecoins are this generation’s Eurodollars, and tokenized U.S. Treasuries are this generation’s money market funds.
Pendle and Hyperliquid as early examples of the next stack
The article says the fixed-rate protocols that failed were trying to build interest-rate markets before real rates existed on-chain. They used token emissions and leverage loops to imitate yield. But a full yield curve cannot be built on returns that can be turned off by governance votes.
Pendle survived, the author argues, because tokenized Treasuries and yield-bearing stablecoins eventually arrived, giving the market access to real cash flow backed by government debt. On that foundation, products became viable. The article adds that Pendle’s Boros is now the first on-chain product in the interest-rate swap category that can operate normally.
The same shift, it says, is showing up in trading infrastructure. In liquid trading pairs, AMM liquidity providers lose about 11% of capital each year to arbitrage leakage. For tokens not listed on centralized exchanges, bearing that "arbitrage tax" was once the only option. For assets such as Treasuries, equities and bonds, which already have deep global liquidity, that model makes much less sense.
Those assets need on-chain order books with performance closer to centralized exchanges. The article notes that many public-chain teams spent years arguing this was technically out of reach, but Hyperliquid challenged that claim. In early 2025, it at one point handled about 60% of on-chain perpetual futures trading volume, according to the piece.
The article closes with a clear direction. Infrastructure built for tokenized Treasuries and tokenized equities is likely to resemble traditional electronic markets far more than many expect, because those assets were already served by traditional finance long before they came on-chain.
The first generation of DeFi components was born in a market dominated by anonymous wallets holding volatile tokens with no cash flow. Now the underlying assets are changing. As that base shifts, the old workaround structures may fade, replaced by infrastructure that looks increasingly similar to the market architecture that has processed trillions of dollars in daily volume for decades.