Pendle as Yield Infrastructure

Summary

In 2025, Pendle shifted from a general yield-tokenization protocol into infrastructure for locking stablecoin yields at fixed rates. ETH derivatives, once Pendle's core use case, declined to just 2.9% of open interest. Activity consolidated around a narrow set of stablecoin instruments where implied rates trade persistently at 8–20%, well above ETH derivative yields of 3–7%.

Users overwhelmingly buy and hold. 85.6% of capital stays until maturity, 98.9% of wallets never sell, and only 7.4% of expired capital rolls into the next expiry. Pendle operates as a one-way funding lock rather than a trading venue.

But this capital doesn't arrive directly. 73.8% of Pendle's external PT open interest flows through lending protocol intermediaries: Aave V3 alone holds 54.7% across 26 vault contracts. The dominant strategy is looping yield-bearing stablecoins: deposit into Pendle, post PT as collateral on a lending venue, borrow stablecoins, re-enter Pendle. This makes Pendle's scale inseparable from the lending venues that consume its output, and inflates headline open interest.

Key findings:

• $538M in active PT open interest, with 96.1% in stablecoins.
• About 66-day weighted average duration: the average dollar on Pendle is exposed to rate moves for roughly two months. Pendle is short-duration yield infrastructure, not a term funding venue.
• 93% of dollar-denominated OI in holding-behavior markets: positions are sticky, not churning.
• 54.7% of all PT units are held by Aave V3: Pendle is an intermediary-consumed infrastructure and integrated directly with lending venues, not a direct-to-user venue.
• We estimate 10–31% of OI is leverage-generated rather than independently committed capital.
• Only 7.4% of expired capital rolls into the next expiry. When PT positions expire without rollover, the associated lending positions (loops, collateral commitments) must also unwind, creating periodic deleveraging pressure across the lending venues that warehouse PT.
• 66.1% of outstanding YT sits with addresses that only ever minted (PT-motivated), while just 15.5% represents independent yield-long demand. The yield side of Pendle is largely a byproduct of principal token manufacturing, not the primary product.

What Pendle Is

Pendle is infrastructure for separating, pricing, and transferring future yield. It tokenizes yield that originates elsewhere. Pendle takes yield-bearing assets and splits them into their principal and yield components, each tradeable independently through a dedicated AMM. Through the venue participants can lock fixed rates, trade yield exposure, or hold positions until maturity.

The mechanism begins with Standardized Yield (SY), a token wrapper conforming to EIP-5115 that provides a common interface for any yield-bearing asset and standardizes how Pendle interacts with the full range of yield-bearing assets across DeFi. When a token is deposited into Pendle, it is first wrapped into its SY representation (e.g., sUSDe becomes SY-sUSDe).

SY is then split into two tokens in equal quantities: a Principal Token (PT) and a Yield Token (YT). The split is atomic: 1 SY always produces 1 PT and 1 YT, and the two cannot be created independently.

• A Principal Token is a claim on the underlying principal at maturity. PT trades at a discount to its underlying asset, and that discount narrows as maturity approaches, converging toward par at expiry. This predictable price path is what makes PT function as collateral on external lending venues, a dynamic central to how Pendle's open interest is consumed downstream.
• A Yield Token is a claim on all yield generated by the underlying from the point of acquisition until maturity. YT holders can claim accrued yield at any time. At expiry, YT's value converges to zero: whatever yield was generated has been captured, and the token carries no residual claim. After maturity, only PT is required to redeem the underlying.

Each underlying asset can support multiple Pendle markets at different maturities, each with its own liquidity pool and its own implied rate. Trading occurs through a single PT/SY liquidity pool per market. The AMM is time-aware: its curve shifts to account for PT's natural price appreciation as maturity approaches, concentrating liquidity around expected yield ranges and reducing impermanent loss for liquidity providers. YT trades route through the same pool via flash swaps. A single pool serves both sides of the yield split.

SY-Notional in 2025: Activity & Composition

SY-notional captures the priced economic throughput flowing through Pendle's V3 markets: the daily volume of Standardized Yield tokens repriced via the AMM, representing how much yield exposure changes hands.

• For most of 2025, daily swaps ranged 500–2,000 with trader counts tracking closely. Activity spiked sharply in early November: daily swaps exceeded 9,400 and unique traders hit 8,000.
• Late-2025 throughput was dominated by Hyperlend, Ethena (USDe/sUSDe), Resolv, Open Dollar – almost entirely stablecoin-sourced. ETH-derivative swap activity concentrated in H1 2025 and became minimal after June.

ETH-Derivative Throughput

SY-notional across a whitelist of ETH-derivative underlyings (weETH, pufETH, rsETH, rswETH, tETH) shows extreme sparsity punctuated by episodic spikes.

Q1 throughput totaled $90.4M, growing to $116.1M in Q2 before collapsing post-June to $16.2M in Q3. The decline is more severe than USD figures suggest: ETH rallied from ~$2,800 to a $4,900 ATH in late August, meaning constant activity levels would have produced higher Q3 volumes, not lower. The USD-denominated drop understates the real contraction in ETH-notional terms.

PT-notional for the same ETH-derivatives shows expiry-driven spikes in late May and June, likely tied to duration compression and roll/unwind events. Outside these episodes, PT-notional is minimal, consistent with steady-state rolling rather than active rate trading.

Open Interest

Open interest measures total Principal Tokens outstanding: capital committed to fixed-rate positions that hasn't expired or been redeemed. Where SY-notional and PT-notional measure flow (what passed through), OI measures stock (what stayed).

Three lenses apply: PT OI in units (principal committed), PT OI in USD (dollar value of commitments, requiring pricing), and duration outstanding (PT OI × time-to-maturity, yielding rate-dollar-years of risk in the system).

Unit-Denominated Open Interest: Scale & Concentration

Scale & Concentration

Aggregate active PT OI stands at 538M units across 96 unexpired markets. Concentration of current active markets is extreme: the top 5 markets hold 80% of OI, the top 10 hold 95%.

The leading markets (PT-sUSDe-7MAY2026 (41.8%), PT-srUSDe-2APR2026 (15.4%), PT-sNUSD-5MAR2026 (13.8%)) reflect the current expiry cycle with asset concentration around the Ethena ecosystem and Strata's yield products (sUSDe, srUSDe, USDe, sENA, jrUSDe collectively ~65% of active OI).

This is contingent on current market preferences in this particular window of time and could shift with subsequent expiry cycles. Active markets vary greatly as throughout compiling this report some have expired.

Temporal Concentration

Temporal concentration in Pendle markets is structural, reflecting how yield markets work:

• Lifecycle accumulation: older markets have had more time to build OI
• Capital efficiency: traders prefer shorter lockups
• Price certainty: PT converges to par at expiry
• Roll behavior: traders exit and roll into next maturity

Unlike asset concentration (which is contingent), the preference for sub-90-day maturities is structural: PT buyers prefer shorter duration locks for capital efficiency and price certainty (PT converges to par at expiry, reducing mark-to-market risk).

The bulk of active OI sits in the 30–90 day bucket (58%), followed by 0–30 days (14.4%). The 180+ day bucket is essentially empty. This mid-curve concentration reflects the current expiry calendar: the March through May 2026 cluster carries the majority of commitments.

Asset Composition

• Stablecoin dominance is total across all buckets. ETH-derivative categories (eETH, stETH, Other ETH) are essentially minimal beyond the 0-30 day bucket.
• Different stablecoins dominate different parts of the curve:
  • 0–30 days: NUSD (Neutrl) at 92.9%, reflecting near-term maturity concentration in a single issuer
  • 30–90 days: Ethena ecosystem at 97.7% (USDe variants 90.9% and other non-stablecoin assets like sENA 6.9%), representing the core of Pendle's current OI.
  • 90–180 days: cUSD (Cap) at 57.7%, with a broader mix including other USD-denominated positions (28.8%) and a smaller Ethena share (6.9%)

Dollar denominated PT Open Interest

Pricing PT open interest quantifies warehoused funding commitments on Pendle in economic terms. Active dollar-denominated OI stands at $538M across 45 priced markets.

• Stablecoins account for 96.1% ($517M), ETH derivatives 2.9% ($15.7M), and other assets 1.0%. The stablecoin dominance observed in unit terms carries through to dollar terms without distortion
• The pricing confidence breakdown for stablecoins: 47% carries direct observable swap prices, 16% relies on the $1 peg assumption, and 37% uses manual estimates (primarily for yield-bearing wrappers like sUSDe).

Duration outstanding

Duration outstanding measures total rate sensitivity in Pendle's positions. It's the answer to "if yields move by 1%, how many dollar-years of exposure are affected?"

• Total duration outstanding is 98.9M rate-dollar-years against $546M in OI
• The weighted average duration is 0.18 years (66 days), meaning the average dollar on Pendle is exposed to rate moves for roughly two months.
• The distribution across duration buckets reveals a proportional relationship between capital and risk:
  • The majority of capital (70.4%) sits in 30–90 day positions, and those positions carry a roughly proportional share of duration risk (67.2%). This is centered on the March–May 2026 expiry cluster.
  • Near-maturity positions (<30 days) hold 17.3% of OI but contribute only 7.5% of duration risk. The mild inversion appears at the long end: the 12.4% of OI in >90 day positions contributes 25.4% of duration risk.

Pendle functions as a short-to-medium duration venue, not a long-duration funding warehouse. The 66-day weighted average duration, combined with high OI, indicates positioning around specific expiry cycles rather than extended rate commitments.

OI/Turnover Ratio

Does Pendle's $538M in active OI sit or churn? Dividing each market's outstanding PT by its 30-day PT-notional traded produces a stickiness ratio that can tell us how many months of current turnover it would take to cycle through the entire position.

• The seven largest markets (PT-sUSDe-7MAY2026, PT-reUSD-25JUN2026, PT-srUSDe-2APR2026, PT-sNUSD-5MAR2026, PT-sENA-30APR2026, PT-USDe-7MAY2026, and PT-stcUSD-23JUL2026) collectively hold $529M and all classify as high-stickiness warehousing (ratio >2x).
  • PT-sUSDe sits at 107x, meaning its entire 30-day trading volume would need to repeat over a hundred times to equal the outstanding position. PT-USDe shows 60x.
  • Even the "lower" ratios in this group – srUSDe at 5.0x, sNUSD at 3.6x – imply positions measured in months of turnover, not days.
• A handful of mid-sized markets (PT-jrUSDe at $15.7M, PT-savUSD at $13.7M, PT-cUSD at $8.1M) show moderate stickiness with ratios near 1.0–1.6x, where some active positioning coexists with warehousing.

No market with meaningful OI classifies as an active trading venue. Roughly 93% of dollar-denominated OI sits in warehousing-behavior markets. Capital enters, locks a rate, and holds to maturity. The turnover observed is consistent with marginal rebalancing and new entry, not active rate trading or position rotation.

Implied Rate Curve

The implied rate is the annualized yield embedded in Pendle's Principal Token (PT) prices. When a PT trades at a discount to its underlying SY token, that discount implies a yield to maturity. This curve represents the market-priced cost of future yield across different maturities within Pendle, emerging from the AMM pricing of PTs relative to their underlying assets.

Average implied rate across all active Pendle markets, grouped by asset class shows us:

• Stablecoin yields trade at 8-20%, while ETH derivatives remain anchored at 3-7%. This percentage point spread persisted throughout 2025. Note that stablecoin PT rates approximate total return, while ETH derivative PT rates compensate only for yield, the holder also carries the underlying's price risk. The gap reflects different risk compositions, not a simple premium.
• ETH derivatives show less volatility than stablecoin yield as a combined asset class.

An important structural note on maturity design: Pendle does not create short-tenor products. Of 182 historical stablecoin markets, 53% were designed with 91–180 day tenors and only 4% with sub-30 day tenors. ETH derivatives skew even longer – no sub-30 day product has ever existed, and 18% of ETH markets carry 365d+ tenors. This means that when short-dated rates appear in the data, they represent aging products approaching expiry, not purpose-built short-term funding instruments. Stablecoin markets are expanding into longer duration, with the 181–365 day bucket showing 16 active versus only 5 expired markets.

Popular Products by Open Interest

To understand which products drive aggregate rate behavior, we examined implied rates for markets where capital is actually committed.

• Ranking products by current open interest shows the top 10 products by OI are mostly stablecoin-adjacent, no ETH derivative appears. SY-sUSDE leads at $220.6M, followed by SY-reUSD ($85.5M), SY-srUSDe ($81.5M), and SY-sNUSD ($71.3M).
• SY-sENA's ($25.2M OI) average rate of 25.82% with a CV of 34.9% represents a fundamentally different yield profile from core stablecoin funding products.
• Several of the top-10 products (reUSD, jrUSDe, srUSDe, NUSD) only launched mid-2025, meaning their yield statistics cover a shorter observation window than full-year products like sUSDE or USDe.
• High open interest does not imply stable yields. To assess yield predictability, we computed the coefficient of variation for each product, the ratio of rate volatility (standard deviation) to average rate. This normalizes volatility across different yield levels.
  • The largest products show wide dispersion in yield stability: SY-sUSDE and SY-USDe exhibit CV percentages of 43–44.6%, meaning nearly half the average rate level is consumed by volatility.
  • By contrast, products like SY-jrUSDe (CV 12.8%) and SY-NUSD (CV 18%) offer meaningfully tighter yield predictability.

Yields: Stablecoin vs. ETH Derivatives

• Within asset class, products move directionally similar. The two asset classes are presented to characterize each rate regime independently. Cross-class rate comparisons do not reflect risk-adjusted relative value.
• In stablecoins:
  • Ethena products (sUSDE, USDe) are the only two with full-year coverage, compressing from ~20–23% in January. sUSDE stabilized around 8–10% while USDe drifted to 3–5% by year-end.
  • srUSDe, sNUSD, reUSD, stcUSD arrived into a lower-rate environment and traded in the 8–15% range. reUSD stands out as the most volatile of the group, spiking to 20% around July. sNUSD and srUSDe held relatively tighter bands in the 8–13% range.
  • stcUSD (Cap) traded the narrowest band of the six until a sharp drop around October, consistent with its low CV of 22.1%.
• ETH-based products cluster around 2–5% with two notable exceptions: SY-tETH spiked to 17% in February and 20% in June 2025. Outside these spikes, ETH products on Pendle offer lower but more predictable fixed yields.

User Behavior and Capital Concentration on Pendle

This section provides a snapshot of the current state of Pendle, capturing who holds what now and how they behaved up to this point, rather than tracking how these distributions evolved over time.

Hold-to-Maturity Behaviour

• Across all markets with >1M units of open interest (approximately equivalent to >$1M for stablecoin markets), 85.6% of capital acquired via swaps is held toward maturity. This aligns with the OI/Turnover stickiness ratio: Pendle functions as a funding lock venue where participants acquire PT positions.
• Wallet-level passivity is extreme: 98.9% of buyer wallets never sell any portion of their position prior to maturity. Early exits are driven by roughly 1% of participants, likely active managers or LPs rebalancing positions.
• Hold rates show no meaningful correlation with market size. Markets ranging from $3M to $225M OI cluster between 80-100% hold rates, with no systematic decline at larger scale.
• Similarly, time-to-maturity does not materially affect exit behavior. Whether a market has 20 days or 160 days to expiry, hold rates remain clustered between 72-99.6%, with one outlier market (sYUSD) at 41%.

Rollover

Rollover analysis examines whether demand for Pendle yield exposure is persistent or episodic. A high rollover rate would indicate wallets systematically re-entering after expiry, demand from participants for the same markets. A low rollover rate suggests one-time tactical positioning.

• Venue-wide swap-based acquisition rollover stands at 7.4% across $819.3M in expired capital. Only $60.7M (7.4%) that held expired PT positions re-entered the same protocol's next market by expiry. Pendle currently tends to operate primarily as a single-cycle funding lock.
• Variance by market is significant. Rollover rates range from 0.1% (Cap) to 16.4% (Ethena). Ethena's rollover rate accounts for $44.5M in venue-wide rollover capital, while protocols with substantial expired capital bases – Cap at $216.7M, OpenEden at $117.2M – show rollover rates below 1.5%.
• Only Ethena (16.4%) and Infinifi (13.2%) show rollover behavior consistent with repeat structural use.

The low venue-wide rollover rate has implications for the lending venues that warehouse PT collateral. When PT positions expire without rollover, the associated lending positions – loops, collateral pledges – must also unwind.

Capital Concentration by Holder Type

Open interest at the wallet level gives us a general idea of who holds Pendle's duration commitments. The distribution indicates whether Pendle serves as infrastructure for retail users or indicates sophisticated use and allocators.

• Pendle is not a direct-to-user venue. Only 20% of total PT OI sits with EOAs interacting with Pendle directly (181M units across 4,079 wallets, excluding Pendle infrastructure).
• The dominant access point is lending protocol integration. Aave, Morpho, Euler, and Silo collectively hold 66.1% of total PT OI (73.8% of external OI) across 76 contracts, each aggregating potentially thousands of underlying depositors.
• Aave remains the primary channel: 26 contracts holding 54.7% of total PT OI (552M units). Morpho adds 10.4% (104M units) across just 2 vault contracts. Combined, these two protocols represent 65% of total warehoused capital. Euler (0.8%) and Silo (0.2%) contribute at the margin.
• Multisigs represent 4.2% of OI. This can suggest treasury management or institutional allocations accessing Pendle directly rather than through lending aggregators.
• Wallet counts create an illusion of breadth. The 4079 EOAs sound like broad participation, but they hold only 18% of total units. The 76 lending vault contracts holding 66% are the economically significant participants.
• Among direct EOA holders capital concentration remains very high: Gini coefficient of 0.916. Pendle attracts concentrated capital both through aggregated lending protocol exposure and among direct participants. The venue does not exhibit broad retail dispersion. Including multisig wallets alongside EOAs produces a near-identical Gini of 0.917.

Holder Composition Across Asset Classes

• Lending protocol integration of Pendle is entirely a stablecoin and "other" asset phenomenon. No lending protocol holds net PT positions in ETH derivatives. Stablecoin PT is 77.9% Aave, 7.9% Morpho, and only 10.3% EOA. ETH-derivative PT is held entirely by EOA.
• Within lending protocols, asset specialization is pronounced.
  • Aave V3 holds 535M units, of which 78.6% is stablecoin and 21.4% "other."
  • Morpho holds 61% "other" and 39% stablecoin. The other category includes assets like ENA, CRV, and governance-adjacent tokens and routes disproportionately through Morpho and Euler rather than Aave.
• Other also carries 3.5% in unclassified contracts, likely specialized vaults or strategy contracts outside of our coverage.

Yield Tokens: The Other Side of the Split

Every PT minted on Pendle produces a YT as its mechanical counterpart. The split is atomic: 1 SY deposited yields 1 PT and 1 YT and the two cannot be created independently. Total outstanding supply of each is comparable across active markets. But where they end up is entirely different.

• Across 15 active markets with more than 1M units in PT outstanding, YT balances sit almost entirely in individual wallets. In 12 of 15 markets, more than 93% of YT is held by EOAs. The three exceptions – sUSDE (76.6%), iUSD (82.7%), and siUSD (53.8%) – show the remainder in non-lending smart contracts that custody YT on behalf of depositors.
• No yield tokens sit in lending protocol contracts across all 15 active markets. By contrast, PT distribution varies dramatically depending on whether the underlying has been accepted as collateral on a lending venue. PTs that have been onboarded 59–99% of PT outstanding sits in lending protocol contracts, with the highest concentration in USDe (99.5%) and srUSDe (95.3%).

When a participant splits SY to post PT as collateral on Aave or Morpho, the PT moves into a lending contract. The YT stays in the minter's wallet or is sold to another individual on the secondary market.

In principle, buying YT is a directional bet on yield: a participant who believes realized yield will exceed the rate implied by current PT pricing can buy YT and profit from the difference. YT offers capital-efficient exposure because its price reflects only the yield component, creating leveraged exposure without margin or liquidation risk. We can classify current YT holders by acquisition history: only received YT from the zero address (a mint) versus only from other addresses (secondary market acquisition).

Stablecoin YT across current Ethereum active markets accounts for approximately 524M units across 2,261 holders:

• 66.1% sits with addresses that only ever received YT through minting. This is consistent with the interpretation that the dominant flow is PT-motivated, where YT remains as a byproduct.
• 18.4% with addresses that both minted and subsequently acquired additional YT through secondary markets. These are addresses that manufactured the split, but also came back to increase yield exposure.
• 15.5% with addresses that never minted at all. This represents genuine yield-long demand from participants who went to Pendle specifically to acquire yield exposure without ever manufacturing the split themselves.

The majority of YT is downstream of PT manufacturing. But a meaningful minority (1/3) reflects independent or reinforced yield-long positioning.

The absence of YT from lending protocol collateral markets reflects structural incompatibility with how lending venues assess collateral. PT's collateral profile converges to par at maturity, allowing lending protocols to set high LTVs. YT decays to zero at maturity and collateral value deteriorates over time regardless of market conditions. In cases where the underlying asset experiences losses, YT absorbs those losses first and can theoretically become negative in value. A lending protocol accepting YT as collateral would require either aggressive LTVs that make borrowing uneconomical or continuous monitoring. And while YT trades through the same AMM pool via flash swaps, the execution path is more complex, involving mint/redeem operations alongside swaps, and liquidation of a decaying asset at a predictable price remains harder than liquidating PT.

PT Reuse Across Lending Venues

Pendle is part of the stack available to asset managers as a tool to offer higher APY for their vaults and serves users with the ability to acquire more units of an asset per dollar spent. Acquiring Principal Tokens means buying the underlying at a discount to par and locking a fixed rate for a defined period, forgoing the variable yield in exchange for price certainty. What makes this trade worthwhile depends on the underlying.

• For volatile assets, the discount can create a meaningful spread over alternatives. If a PT-stETH's implied yield exceeds the staking rate available, the buyer can capture the difference. This can also amplify if the underlying asset appreciates. A buyer with high conviction that a volatile asset will appreciate by maturity may prefer raw exposure over a predictable income stream, treating the PT as a discounted directional position.
• For stablecoins, the logic differs. The asset will not appreciate, so buying and holding stablecoin PTs delivers a better or worse rate than holding the underlying directly. The discount alone is not large enough to be a compelling strategy on its own. The only way to meaningfully amplify the fixed-rate return is through leverage.

Two observations emerge. First, in our venue analysis we observed ETH derivatives make up a small portion of Pendle's flows. The locked APY available for ETH derivatives does not create a meaningful spread over alternatives such as staking, generating little demand for PT-stETH and other ETH-linked products. Meanwhile, stablecoin rates are higher and capital deployed is far larger, making the fixed-rate locking decision more attractive. Second, because the dominant strategy for extracting value from stablecoins relies on looping, and we know Pendle's open interest is mostly concentrated on stablecoins: large parts of the OI could be reflexive. A portion of the capital warehoused in Pendle may itself have been generated by borrowing against previously warehoused Pendle positions.

Looping with Pendle

Looping with Pendle works as follows: a participant deposits into Pendle to receive PT, posts that PT as collateral on a lending protocol, borrows stablecoins against it, and re-enters Pendle with the borrowed capital.

Why PTs? As maturity approaches, the discount shrinks toward zero and the collateral appreciates relative to the debt, meaning health factors on the loan improve over time. In stable market conditions, the loan becomes healthier as it ages. PT looping does not make a position liquidation-proof. It reduces liquidation risk compared to looping with volatile collateral like ETH, but the risk remains and arrives from several directions.

• Borrow rate spikes. If the lending protocol's borrow rate rises above the PT's locked-in yield, the loop becomes negative carry. Governance changes to LTV ratios, liquidation thresholds, or delisting of PT collateral markets can also break the loop entirely.
• Depegging of the underlying. A shallow depeg can be offset by the PT's gradual appreciation. However, a rapid depeg can push leveraged positions toward liquidation.
• PT liquidity within Pendle. Liquidity for a specific PT depends on LPs depositing PT and SY into that market's pool. Before maturity, unwinding depends on pool depth. Market stress can cause LPs to withdraw, and new pools with attractive incentives can pull liquidity away from existing markets, draining the exit path that leveraged holders depend on.
• Oracle and infrastructure risk. Oracle malfunction or compromise at any point in the asset stack can force unwinding regardless of position's health.

Participants building leveraged loops through Pendle have strong incentives to stay short-dated: less price sensitivity to yield spikes means less liquidation risk, less time for borrow rates to shift or governance changes to take effect, and faster access to capital for redeployment.

Estimating Leverage Capacity

The yield routing originates in Pendle's AMM but executes entirely in external lending markets. The scope and parameters of available markets define the strategy's reach. Currently three venues are available on Ethereum: Aave, Morpho, and Euler.

• On Aave, two separate parameters govern leverage: Max LTV and Liquidation Threshold (LT), where Max LTV < LT < 1. Users can borrow up to LTV of their collateral value but are not liquidated until LT.
• On Morpho, a single parameter governs both: LLTV (Liquidation Loan-to-Value).
• On Euler, the deployer sets a borrowing LTV and a separate liquidation LTV for each vault-to-vault relationship.

We can estimate leverage capacity using available markets for each venue.

Venue: Aave

Market: PT-sUSDe Stablecoins eMode

Market: PT-srUSDe Stablecoins eMode:

Market: PT-srUSDe USDe eMode:

Venue: Morpho

LLTV serves as both the maximum borrowing limit and the liquidation trigger. The following table shows achievable leverage with the mid-to-high allow-list of permitted LLTV tiers.

Currently, you can find markets along this range. PT-sNUSD-5MAR2026/USDC lists an LLTV of 77% while some markets like PT-stcUSD-23JUL2026/USDC show an LLTV 91.5%.

Venue: Euler

LTV and LLTV values vary vault to vault, and Euler holds more borrow markets than Aave.

Market: PT-jrUSDe / USDC

Market: PT-srUSDe / USDC

Market: PT-srUSDe / borrow eUSDe

For a more practical estimate, we can observe what leverage operating at 70% and 80% of the Liquidation Threshold achieves.

Estimating the reflexive share of open interest

Not all PT open interest represents independent capital. Each looping iteration adds PT to the system without adding new external capital, inflating OI relative to the actual capital committed. 73.8% of PT held outside Pendle's own contracts sits in lending protocol addresses, approximately $356M of the $538M total OI. We cannot observe how much of it is actively collateralizing looped positions versus sitting as collateral for unrelated borrows, but we can bound the estimate.

We vary two assumptions: the share of lending-protocol PT that is actually looped (30%, 50%, 70%) and the leverage of those loops (2x, 3x). These are conservative estimates given the lending parameters documented in the previous section allow for greater leverage.

Taking the most conservative estimate, we assume 30% ($107M) is actively collateralizing looped positions. The remaining 70% ($249M) is either collateralizing unrelated borrows or not borrowed against at all. At 2x effective leverage, each dollar of original capital produces two dollars of PT through one full loop: $107M in looped PT was generated by $53M of original capital. Against total OI of $538M, that $53M in reflexive PT represents a 10% reflexive share. The independent capital actually committed to Pendle is $485M.

The range spans 10-31%. Even the most conservative assumption (30% looped at 2x) implies roughly one in ten dollars of headline OI is leverage-generated. The aggressive end (70% at 3x) implies nearly one in three.

Lending Venue Landscape

Available markets concentrate around a narrow set of venues and underlyings.

On Aave V3, PT collateral acceptance is narrow and concentrated in PT-sUSDe across two markets and, recently, PT-srUSDe. On Morpho, PT collateral markets are broader. Higher-deposit markets include Resupply's reUSD, Strata's srUSDe and jrUSDe, Cap's cUSD and stcUSD, Neutrl's NUSD, Resolv's RLP, Midas' mAPOLLO, InfiniFi's iUSD and siUSD, and Ethena's sUSDe. Most other PT collateral markets remain below $1M in deposits. On Euler, PT markets include Strata's srUSDe, jrUSDe, and pUSDe, Cap's cUSD and stcUSD, Falcon's USDf and sUSDf, sBOLD, Kelp DAO's hgETH, and Midas' mAPOLLO and mMEV.

Who Controls What for Borrow Markets

Aave

PTs are not listed as general collateral but in a dedicated eMode category with specialized pricing. eMode are groupings of correlated assets that hold the same category-specific oracles, stabilizing pricing within these groups and reducing unnecessary liquidations from minor price deviations. By allowing higher LTVs on correlated asset pairs, eMode enables the leveraged yield strategies that generate significantly more borrow demand than standard borrowing. Initial listings require governance proposals and follow a full governance cycle including onchain votes through vote.onaave.com (recent examples include Strata's srUSDe and PT Ethena May) Regardless of past acceptance, each new expiry requires explicit whitelisting.

LTV and risk parameters can be adjusted through the RiskSteward, which updates eMode collateral parameters such as Liquidation Threshold, LTV, and Liquidation Bonus within DAO-approved bounds, while also allowing changes to the discountRate of Pendle PT feeds. In practice, markets are whitelisted with recommended parameters and the RiskSteward adjusts as needed. As maturity approaches, the Edge AGRS system manages risk parameters; if the Pendle AMM reaches 96% PT liquidity concentration, LTV is set to zero and lending is paused.

Borrow rates are determined by the protocol's utilization-based kink model, governed at the DAO level.

PT supply as collateral is capped to reduce protocol exposure and manage concentration risk.

Morpho

Creating new borrow markets for any asset in Morpho Blue markets is permissionless, requiring five parameters: loan token, collateral token, oracle, LLTV, and IRM.

LTV and risk parameters are immutable once created. Governance cannot modify them, though Morpho maintains an allow-list of permitted LLTV tiers and IRM contracts. For PT collateral markets, the most commonly used tiers fall in the 86-94.5% range.

All markets currently use the same AdaptiveCurveIRM, which autonomously adjusts rates based on utilization targeting approximately 90%. No governance or curator can intervene in rate determination.

Euler

Every Euler Vault Kit (EVK) vault is simultaneously a deposit destination and a borrowing source. Unlike Morpho, there is no separate base-layer market underneath a vault layer, EVK unifies both functions. Anyone can list new markets.

LTV and risk parameters are set vault-to-vault, not asset-to-asset. The governor decides which collateral vaults to accept and assigns each a borrowing LTV and a separate liquidation LTV. Depending on vault type, parameters and borrow caps can be changed by the governor. Borrow rates are determined by an IRM contract attached to the vault; the governor selects the IRM and can change it.

Collateral Selection and Underlying Risk

For non-yield-bearing underlyings such as USDe, NUSD, or cUSD, the PT still trades at a discount to par, but the discount does not compensate the holder for foregone yield, the underlying generates none. Instead, the YT side of these markets is priced almost entirely by points programs: Ethena's sats multipliers for USDe deposits on Pendle, Plasma's XPL program, and similar mechanisms.

For yield-bearing underlyings such as sUSDe or srUSDe, the PT embeds a genuine fixed-rate claim. The buyer foregoes variable yield in exchange for rate certainty over a defined period, and the resulting PT is the core substrate for the looping strategies documented in the previous sections. Even without points or multipliers, the spread creates demand. Strata's srUSDe and jrUSDe represent a third category: a fixed-rate claim on the yield of another product. The underlying is itself a redistribution of Ethena's sUSDe strategy through Strata's tranche mechanism. The economic function is the same as the yield-bearing case, but the dependency chain between the PT holder and the original yield source is longer.

Supply-Side Selection

Across Aave, Morpho, and Euler, lending venue listings cluster around a narrow set of underlyings. Several structural filters operate simultaneously to consider which assets get listed on Aave and generate enough interest to be listed on Morpho and Euler.

• Redemption predictability is the first filter. Lending venues need confidence that PT-to-underlying conversion at maturity will execute reliably. The redemption path may depend on coverage ratio thresholds, minting pauses, or withdrawal queues. These dependencies make the underlying's stability over the PT's life the binding constraint.
• Collateral-debt correlation. When you borrow against collateral, the lending venue cares about how likely the collateral's value is to drop below your debt. The closer they move together, the lending venue can set a higher LTV without opening paths for unhealthy loans and leverage.
• Track record. Protocols that lack established history can route through permissionless venues like Morpho or Euler, where the curator layer performs the filtering function instead of full governance cycles.
• Liquidity depth in the Pendle AMM for that specific market. Liquidation of a PT collateral position requires selling PT into the pool. If a market is thin, the lending venue cannot rely on liquidation execution at reasonable prices. Duration of that specific market also determines exposure to risk. From the lending venue's perspective, shorter-dated PT markets are safer collateral.

Demand-Side Reinforcement

Two conditions channel demand toward specific underlyings.

• First, a yield spread over alternatives: the implied rate on the underlying persistently exceeds what a user can earn from holding directly.
• Second, reliable loopability: the Pendle market has depth, a lending venue accepts the PT with a high enough LTV to make the loop worthwhile, and the borrow market has sufficient liquidity to fund re-entry.

Strata's srUSDe listing illustrates the selection framework in action. The asset is stablecoin-denominated, so PT-srUSDe collateralizing a stablecoin borrow is a correlated pair. Redemption follows a defined path: srUSDe redeems to sUSDe, which redeems to USDe, with published coverage ratio thresholds governing each step. Liquidity depth is met through Pendle's AMM, where 97%+ of srUSDe supply sits in the SY-wrapper for PT markets. Despite the protocol being live for four months, srUSDe markets have been used on Morpho and Euler, Strata has built reputation through it's backing by the Ethena Foundation and investments by Anchorage Digital Ventures. From the demand side, the implied rate and LTVs in Aave make increasing leverage attractive.

Pendle's scale cannot be read independently of the downstream lending venues and their parameter decisions that determine what gets listed, leveraged, and looped. That interdependence is what makes Pendle legible as infrastructure rather than as an isolated venue.

---

Methodology

Scope

This report is a descriptive, data-driven assessment of Pendle as yield-tokenization infrastructure on Ethereum in 2025. The analysis frames Pendle as a rates and duration market, as infrastructure for pricing, transferring, and warehousing future yield, and examines how its output is consumed by downstream lending protocols through looping and collateral posting.

Outside of scope:

• PENDLE governance token economics, staking mechanics, or protocol fee distribution.
• LP profitability or impermanent loss analysis for Pendle AMM liquidity providers.
• Comprehensive audit of every PT collateral market across all lending venues; lending venue coverage is limited to the dominant intermediaries (Aave V3, Morpho, Euler) with representative examples.
• Pendle deployments on chains other than Ethereum (Arbitrum, Plasma, Optimism, and others are not covered).
• YT trading strategies or yield speculation profit-and-loss.
• Historical protocol versions prior to Pendle V3, except where needed for context.
• Stablecoin or token design assessment or prescriptive recommendations for token issuers; stablecoins are treated as structural inputs, not operational products.

Definitions and Interpretations

• Standardized Yield (SY): A token wrapper conforming to EIP-5115 that provides a common interface for any yield-bearing asset. When a token is deposited into Pendle, it is first wrapped into its SY representation. SY standardizes how Pendle interacts across DeFi.
• Principal Token (PT): A claim on the underlying principal, redeemable at par at maturity. Trades at a discount to SY that narrows as maturity approaches.
• Yield Token (YT): A claim on all yield generated by the underlying from acquisition until maturity. YT holders can claim accrued yield at any time. At expiry, YT converges to zero, carrying no residual claim.
• Flash swaps: The routing mechanism through which YT is traded via the PT/SY pool. The AMM synthesizes YT exposure by flash-minting SY into PT + YT and selling the unwanted PT in a single atomic transaction.
• Maturity / expiry: The date at which PT becomes redeemable 1:1 for its underlying asset.
• SY-notional: The daily volume of Standardized Yield tokens repriced via the AMM. Interpreted as priced economic throughput (how much yield exposure changes hands) rather than as "volume" in the spot-DEX sense. Sourced from swap events on Pendle V3 markets.
• PT-notional: The volume of Principal Tokens traded through the AMM. Interpreted as duration and funding turnover: the primary signal of rates trading behavior. PT-notional diverging from SY-notional may indicate shifts in the balance between yield transfer and rate positioning.
• Open interest (OI): Total Principal Tokens across active (unexpired) markets (capital committed to unredeemed fixed-rate positions). Expired markets retain OI until redeemed but are excluded from active counts. OI is reported in three forms:
  • Unit-denominated OI: PT units outstanding. Computed as net supply from erc20_ethereum.evt_Transfer for each PT token address with market metadata, reconstructed from CreateNewMarket events across all three factory contracts. Underlying type categorized via symbol pattern-matching against the tokens.erc20 table.
  • Dollar-denominated OI: PT OI valued in USD. A VWAP PT price per market is derived from Pendle swap events over the trailing 15 days, computed as the ratio of SY-side notional to PT-side notional; markets without recent swaps default to a 0.95 PT price. An underlying USD price is sourced from prices.usd_latest on Dune, matched by stripping the "SY-" prefix from the token symbol; tokens absent from that feed receive manually assigned USD prices (e.g., 1.0 for stablecoin underlyings, 1.14 for sNUSD/sRUSDe). Final dollar OI per market equals oi_units x pt_price x underlying_usd_price, aggregated by asset class.
  • Caveats: OI figures exclude a legacy PT-stETH-30DEC2027 position minted by a legacy contract factory, representing <0.1% of active OI.
• Duration outstanding: PT OI multiplied by time-to-maturity for each position, yielding rate-dollar-years of risk in the system. This measures total rate sensitivity, or how many dollar-years of exposure are affected by a 1% yield move.
  • Computed as the sum of (PT OI in USD multiplied by time-to-maturity in years) across all active markets. Market metadata sourced as described in OI methodology above, joined with PT supply and pricing data.
• Weighted average duration: Total duration outstanding divided by total dollar-denominated OI. Represents the average time horizon of a dollar on Pendle: how long the average position is exposed to rate moves. Computed in the same query as duration outstanding.
• OI/Turnover ratio: A market's outstanding PT OI divided by its trailing 30-day PT-notional traded. Ratios above 2x indicate warehousing behavior (positions held, not actively traded); ratios near 1x indicate active trading with regular turnover.
  • Numerator is all-time net PT supply from erc20_ethereum.evt_Transfer, filtered to positive balances and unexpired markets. Denominator is trailing 30-day PT-notional from swap events. Both joined via market metadata as described above.
  • Markets are classified by ratio: >2x (warehousing), 0.5-2x (positioning + trading), 0.1-0.5x (repricing venue), and <0.1x (transient flow).
• Hold-to-maturity rate: The share of capital acquired via swaps that is held toward maturity without being sold prior. These figures represent a conservative lower bound on actual hold-to-maturity behavior. The analysis captures only wallets that acquired PT through AMM swaps. Computed at the wallet level: for each wallet that acquired PT through a swap, track whether any PT was transferred out (excluding burns/redemptions) before the market's expiry date. Sourced from wallet-level erc20_ethereum.evt_Transfer analysis for PT tokens, filtered to swap-acquired positions.
• Rollover rate: The share of expired capital that re-enters the same protocol's next maturity market by expiry. Sourced from cross-market wallet tracking via erc20_ethereum.evt_Transfer, joining expired and active markets for the same SY underlying, divided by total expired capital for that underlying.
• Gini coefficient: Measures how concentrated PT holdings are among direct EOA holders, from 0 (evenly spread) to 1 (held by one wallet). Computed from PT holdings across EOAs, excluding lending contracts, multisigs, and Pendle infrastructure.
• Implied rate: The annualized yield embedded in Pendle's PT discount to SY. Treated as a forward funding curve (the market's expectation of annualized yield over the remaining term) rather than a simple APY. Sourced from UpdateImpliedRate events and daily OHLC data. Rate extraction: bytearray_to_uint256(data) * 100 / 1e18. Implied rates reflect market expectations, not guaranteed returns. See Pendle Docs - PT Pricing.
  • Aggregation: Daily close rate per market (last rate of each day), averaged across all active markets within each asset class.
  • Expiry noise excluded: Filtered out rates where days_to_expiry = 1 to avoid annualization distortion.
  • Asset classification is heuristic, based on SY symbol pattern matching (e.g., "USD" leads to stablecoin). Some edge cases may be misclassified.
• Coefficient of variation (CV): Standard deviation of implied rates divided by the mean, measuring yield stability across products. Comparisons across products with different observation periods (full-year vs. mid-2025 launches) should be interpreted with caution. Computed from the daily implied rate time series.
• Dollar-denominated OI pricing methodology: Not all PT positions carry equally observable prices. Three pricing tiers apply to stablecoin OI:
  • Direct observable swap prices: VWAP from recent swap events, as described in OI methodology above.
  • $1 peg assumption: Non-yield-bearing stablecoin underlyings (e.g., cUSD) where the PT claim is on a dollar-pegged asset.
  • Manual estimates: Yield-bearing wrappers (e.g., sUSDe, srUSDe) where no direct swap price exists, valued via the wrapper's known exchange rate.
• ETH-derivative pricing uses 24-hour average WETH price from prices.usd on Dune.
• Max LTV: Defines the maximum ratio of borrow value to collateral value at origination.
• Liquidation Thresholds (LT): Liquidation Threshold defines the ratio at which positions become eligible for liquidation. LT is always higher than Max LTV, creating a built-in safety buffer for venues that have both metrics. In this report we referenced two venues that do Aave (Aave V3 Documentation) and Euler, see below for their definition.
• Borrowing LTV and Liquidation LTV: Euler's vault-to-vault model. Each collateral-borrow vault pair has a Borrowing LTV (maximum borrow ratio at origination) and a Liquidation LTV (ratio triggering liquidation). Structurally similar to Aave's two-tier system, but governed per-vault, with parameters adjustable by vault governors. See Euler V2 Documentation and the Euler app for live vault configurations.
• LLTV: Liquidation LTV (LLTV) defines the collateral ratio at which a position can be liquidated. Unlike Aave, in Morpho's single-parameter model there is no separate origination LTV: borrowers can borrow up to any ratio below LLTV at their own discretion. LLTV is set at market creation and is immutable for the life of that market. See Morpho Docs - Creating a Market and Morpho Blue contract source.
• Reflexive OI / leverage-generated OI: The portion of open interest created by looping rather than independent capital commitment. Estimated as a bounded range. Actual on-chain leverage cannot be directly observed from PT holdings alone without wallet-to-wallet analysis.
• Independent capital: Total OI minus the estimated reflexive share.
• EOA (externally owned account): A standard Ethereum wallet controlled by a private key, as distinguished from a smart contract. EOA holders represent direct, non-intermediated participation. Identified by checking against ethereum.creation_traces.
• External OI: PT open interest held outside of Pendle's own infrastructure contracts (routers, AMM pools, factory contracts) and third-party aggregators (Penpie, Equilibria).
• Mint-only holders: Addresses that received YT exclusively from the zero address (through SY to PT + YT minting). Interpreted as PT-motivated minters. Identified via erc20_ethereum.evt_Transfer analysis for YT tokens.
• Secondary-market-only holders: Addresses holding YT acquired entirely through market purchases or transfers, with no minting activity. Interpreted in the report as genuine yield-long demand.
• Stablecoin vs. ETH derivative vs. Other: The taxonomy for classifying Pendle market underlyings by economic exposure. Stablecoins include dollar-pegged and dollar-yield-bearing assets; ETH derivatives include liquid staking and restaking tokens (weETH, pufETH, rsETH, rswETH, tETH); Other captures assets not fitting either category. Classification applied via SY symbol pattern matching, consistent across all markets.

Data Sources

• Pendle protocol documentation: For protocol mechanics, SY/PT/YT definitions, EIP-5115 compliance, AMM design, flash swap routing, and redemption mechanics. docs.pendle.finance. Pendle infrastructure and factory contracts were taken from the protocol's GitHub.
• Queries: PT Discovery and every query that links PTs to markets (for VWAP pricing, turnover, or swap data) all draw from factory contracts. Unit OI, dollar OI, concentration, expiry buckets, asset composition, duration, and OI/turnover all draw from the same factory pair + expiry filter. Throughput queries (SY-notional, PT-notional) are independent, they pull from decoded swap events. Swap events are decoded from a topic in ethereum.logs, where netPtOut (bytes 1-32) and netSyOut (bytes 33-64) capture trade sizes. Transfer analysis uses erc20_ethereum.evt_Transfer for PT and YT token balances, mints, burns, and holder tracking. Contract classification uses ethereum.creation_traces to distinguish EOAs from smart contracts and to identify deployer relationships. All queries are available in our Dune profile.
• EIP-5115 specification: For the Standardized Yield token standard that underpins Pendle's SY wrapper. See EIP-5115.
• Lending venue documentation and governance:
  • Aave: Governance proposals via vote.onaave.com, specifically Proposal 299 (PT-sUSDe and PT-eUSDe onboarding, Edge AGRS activation) and Proposal 444 (Strata PT-srUSDe onboarding). Aave community forum discussions for risk parameter context. Aave V3 technical documentation for eMode mechanics, interest rate models, and risk management frameworks.
  • Morpho: docs.morpho.org for market creation mechanics, LLTV tier governance, and AdaptiveCurveIRM design. Morpho Blue contract source code for enableLltv immutability verification.
  • Euler: docs.euler.finance for Euler Vault Kit architecture, vault governor mechanics, and LTV parameter structures. Euler app interfaces for live vault configurations and collateral parameters.
  • LlamaRisk: Governance forum analyses including LlamaRisk Insights: Aave's PT Token Exposure Risk Outlook and risk assessments in the ARFC for Strata srUSDe onboarding.
  • Chaos Labs: Edge Risk Oracle documentation and risk parameter recommendations cited in Aave governance proposals.
• Protocol applications and UIs: Morpho app (app.morpho.org), Euler app (app.euler.finance), and Aave UI for observing live market parameters (LTV, LLTV, supply caps, utilization) at the time of report compilation (February 3rd, 2026).

Known Limitations

• Snapshot timing: OI, holder composition, and rate data are point-in-time snapshots. Active markets change as positions expire and new markets launch. All queries mentioning "current on-chain data" correspond to February 9th and February 10th.
• Observation windows: Products launched in H1 2025 carry full-year observation periods for implied rates and throughput. Products launched mid-2025 have shorter windows. CV and average rate comparisons across products with different observation periods should be interpreted accordingly.
• Dune indexing: All quantitative analysis depends on data indexed and available through Dune Analytics at the time of compilation. Indexing gaps, delays, or reorgs may affect completeness. Event decoding relies on known contract ABIs and topic hashes.
• Chain coverage: Analysis is Ethereum-focused. Pendle operates on Arbitrum, BNB Chain, Optimism, and other chains, which are not covered.