Pendle peaked at $8.9 billion in TVL in August 2025. Today the dashboard reads $1.539 billion. The panda has stared at the chart, and the chart has stared back.
That is not a death. That is a chapter change. The LRT wave receded, the EigenLayer points loop dried up, and Pendle was left holding the only piece of DeFi infrastructure that mattered through the cycle: a venue where future yield gets a price. The question is whether the next chapter, Boros and RWAs, actually fills the hole.
From $8.9B to $1.5B: What Just Happened
According to DefiLlama's Pendle protocol page, Pendle's TVL has fallen roughly 83% from its August 2025 peak. The bulk of that drop tracks the broader retreat in liquid restaking deposits, which was Pendle's biggest pool of underlying assets through 2024-2025. When the points speculation cooled, the YT bid cooled with it. As always in crypto, the narrative leaves and the plumbing stays.
The broader DeFi context is not collapsing in tandem. Total DeFi TVL sits at $80.73 billion across all chains, with Ethereum alone at $42.27 billion. The aggregate is roughly stable. The composition is what moved, away from speculative restaking yield and toward stablecoin yield and (slowly) tokenized real-world assets.
Pendle's residual TVL tells the rest of the story. Per industry reporting, Ethena's USDe represents roughly 70% of Pendle's current pool exposure. The protocol stopped being a points casino and quietly became a yield-curve venue for synthetic dollars. Less hype, more cash flow.
How does Pendle actually split yield?
Pendle takes any yield-bearing asset (stETH, sUSDe, GLP, whatever) and splits it into two tokens. The PT (Principal Token) represents principal redeemable 1:1 for the underlying at maturity. The YT (Yield Token) represents all the variable yield that asset will accrue between now and maturity. PT plus YT equals the original position.
The mechanism is the part most articles skip, which is unfortunate because it is the part that makes Pendle interesting. Buying PT at a discount and holding to maturity locks in a fixed rate, because the discount converges to par. Buying YT is the opposite trade: you pay upfront for variable yield, profitable only if realized yield exceeds the implied rate priced into YT. According to Pendle's official documentation, the system is effectively a zero-coupon bond market for on-chain assets, with the AMM doing price discovery.
Why does this matter beyond yield-farming theatre? Because every DeFi yield (restaking, stablecoin, RWA, lending) has a number attached to it that nobody really prices in advance. Pendle gives that number a market. Everything else (Boros, RWA vaults, the Spark stUSDS integration) is the same mechanic applied to fresh collateral.
Boros: Funding Rates as a Tradable Curve
Pendle's August 2025 launch of Boros extends the yield-splitting machinery to a market most retail traders only feel indirectly: perpetual futures funding rates. The daily funding-rate flow across crypto perps is estimated at $150 to $200 billion in notional. Until Boros, there was no clean on-chain way to either hedge that exposure or speculate on it.
Boros introduces Yield Units (YUs). One YU corresponds to the funding stream earned or paid on one unit of an underlying perp position until maturity. So a trader can buy or sell future funding without taking the underlying directional bet. According to OAK Research, this turns a side-effect of perpetual contracts into its own tradable instrument, the closest thing DeFi has yet built to a funding-rate future.
For context on the size of the prize, BTC trades at $73,780 today and ETH at $2,020 (data from CoinGecko's bitcoin page). On those notionals, even a 5 to 10% annualized funding swing translates to seven-figure PnL on a single basis trade. That is the addressable market Boros is shooting at. Whether on-chain liquidity catches up to centralized perp venues fast enough to capture meaningful share is the real question, not the elegance of the design.
The Pendle team has published a cross-exchange funding-rate arbitrage strategy using Boros, which is more or less the textbook basis trade dressed up for on-chain settlement. Useful. Also mildly suspicious in the way every "fixed yield" pitch is mildly suspicious until you read the risk section.
Risks: Smart-Contract, Oracle, and Curve Misalignment
Smart-contract risk is the obvious one. Pendle has been audited multiple times and has not been exploited at protocol scale, but the system is unusually composable: AMM, PT/YT splitting, Boros's funding settlement, oracle hookups for sUSDe and similar collateral. Each integration is a new surface. Composability is also Pendle's biggest attack vector.
Oracle and depeg risk is the underrated one. Pendle's PT redemption assumes the underlying yield-bearing asset converges to par at maturity. If sUSDe depegs, or if stETH suffers a slashing event, the PT no longer redeems 1:1, and what looked like "fixed yield" was actually short volatility on the collateral. With USDe-linked pools dominating current TVL per the same DefiLlama dashboard, concentrated collateral means concentrated risk. The numbers say yes. The panda raises an eyebrow.
Yield-curve misalignment risk is the niche one. If Pendle's implied YT rate prices in funding or staking yield that does not materialize, YT buyers eat the loss. The protocol does not promise anything. It just clears a market. Governance risk and YT exit liquidity before maturity round out the list. Most retail YT buyers do not understand they own a depreciating asset. That is not a Pendle bug. That is a comprehension tax DeFi has always charged.
What to Watch Next
Three signals will decide whether Pendle's pivot works. First, Boros TVL and active YU markets: does funding-rate trading actually move on-chain, or does it stay on Binance and Bybit? Second, the Spark stUSDS vault integration and whether Sky/Maker collateral builds a second leg beyond Ethena. Third, RWA TVL growth: roughly $151M today is a rounding error, but on a longer horizon, fixed-income desks moving on-chain need a yield curve. Pendle already operates the only credible one.
For builders closer to the Dadacoin and BSC stack: yield tokenization is one of the few DeFi primitives that genuinely benefits gaming and consumer apps, because it lets a creator pre-monetize a future revenue stream without dumping tokens at spot. Worth watching whether BSC DeFi infrastructure eventually imports the model. Most won't. The few that do will quietly out-engineer the rest. Spoiler: we saw this one coming.
For deeper context, our explainer on perpetual futures mechanics maps the funding-rate plumbing Boros sits on top of, and our DeFi cluster overview lists adjacent yield primitives worth a read.
The panda watches and judges. Pendle is no longer the loudest protocol on the dashboard. It might be the most load-bearing.
