Author:Zack Pokorny,Galaxy
Compiled by:Saoirse,Foresight News
As the market deleveraging trend continues, Q2 2026 marks the first quarter since Q4 2022 where the scale of on-chain lending across all categories—including centralized finance (CeFi), decentralized finance (DeFi), and the crypto-collateralized portion of Collateralized Debt Position (CDP) stablecoins—has declined synchronously.
A key distinction from the previous bear market is that the outstanding loans have decreased in a gradual, stepwise manner rather than in a single crash. In Q2 2022, the crypto-collateralized lending industry plummeted by over 55%, followed by further declines of 9% and 29% in Q3 and Q4 2022, respectively. In contrast, this deleveraging cycle has seen consecutive quarterly declines of only 10%, 5%, and 17% over the last three quarters.
We believe this milder downward pace signifies a healthier deleveraging cycle, driven by markets actively and gradually reducing risk rather than by large-scale forced liquidations or counterparty failures. Should lending volumes continue to contract, we expect this stair-step decline pattern to persist, avoiding a repeat of the chain-reaction, massive losses seen in 2022.
At the corporate treasury level, deleveraging was primarily driven by Strategy Inc.'s completion of a $15 billion debt buyback in May 2026. This reduced the debt used to support digital asset treasury strategies to $16.1 billion, roughly returning to the debt level for such companies from July 2025.
In the futures market, the overall open interest (OI) at the end of the quarter showed little change, declining only 3.08% quarter-on-quarter (QoQ) to $103.2 billion. Beneath this slight overall decline, there was significant internal divergence: Bitcoin OI fell 6.24% to $45.04 billion, while Ethereum OI dropped more sharply by 26.31% to $21.99 billion. At quarter-end, Bitcoin and Ethereum combined accounted for 65% of total futures open interest. It's worth noting that this stability at quarter-end did not last; by the end of July, futures OI had rebounded to approximately $114 billion, with both Bitcoin (~$48 billion) and Ethereum ($25.74 billion) recovering from their Q2 lows.
Key Takeaways
- Overall, the size of crypto-asset collateralized lending contracted by $11.33 billion (-16.78%) in Q2 2026 to $56.16 billion, down 40.13% from its peak of $78.69 billion in Q3 2025.
- The USD-denominated outstanding loans in DeFi lending applications declined for the third consecutive quarter, decreasing by $7.79 billion (-27.61%) this quarter to $20.43 billion.
- According to Galaxy Research's tracking, the outstanding debt used by corporations to directly purchase or supplement digital asset treasury strategies stands at $16.1 billion.
- Futures open interest (OI), including perpetual contracts, declined 3.08% QoQ to $103.2 billion.
Crypto-Asset Collateralized Lending
The market overview chart below shows the historical and current major players in the CeFi and DeFi crypto lending space. Affected by sharp declines in crypto asset prices and liquidity drying up, several of the largest CeFi lending institutions collapsed between 2022-2023, marked in red on the chart.
Centralized Finance (CeFi)
The table below compares the various CeFi lending institutions covered in this market analysis. Some institutions offer diverse services to investors; for example, Coinbase is primarily an exchange but also provides credit to users through over-the-counter (OTC) crypto loans and margin financing. This analysis only accounts for the balance sheet size of crypto-collateralized lending for each entity.
As of June 30, Galaxy Research statistics show CeFi outstanding loan volume at $22.98 billion, down 9.62% QoQ (a decrease of $2.45 billion). Compared to the Q4 2023 bear market low of $6.8 billion, this represents growth of $16.14 billion, or 235.94%; however, it remains 37.16% below the all-time high of $36.58 billion in Q1 2022.
The contraction in total CeFi lending in Q2 was primarily driven by a decline in Tether-guaranteed outstanding loans; the lending book sizes of Galaxy, Coinbase, Ledn, Arch, Sygnum, and Milo all grew during the quarter.
Tether remains the absolute leader in the CeFi lending market, with a 58.54% market share, down 371 basis points (bps) QoQ. Combined with Maple (8.91%, up 52 bps QoQ) and Nexo (7.51%, up 49 bps QoQ), the top three institutions within our coverage account for 74.96% of the market, with their combined share declining 270 bps QoQ overall.
When comparing market shares, it's important to note significant differences among CeFi institutions: some only offer specific loan types (e.g., BTC-only collateral, altcoin-only collateral, or cash loans in fiat rather than stablecoins); some serve specific client segments (institutional/retail); and operations are subject to jurisdictional limitations. These factors result in varying expansion capabilities among different players.
The table below details Galaxy Research's data sources for each CeFi institution and the logic for calculating their book sizes. Data for DeFi and on-chain CeFi can be obtained transparently and easily from public on-chain data; however, acquiring CeFi data is highly challenging: accounting standards for outstanding loans vary by institution, disclosure frequencies are inconsistent, and comprehensive data sources are difficult to obtain.
Note: Data provided by private third-party sources have not been formally verified by Galaxy Research.
Centralized Finance and Decentralized Finance Lending
The USD-denominated size of outstanding loans in DeFi lending applications declined for the third consecutive quarter in Q2, decreasing by $7.79 billion (-27.61%) to $20.43 billion.
Combining DeFi applications and CeFi lending platforms, the total crypto-collateralized outstanding loans at quarter-end were $43.41 billion, down $10.24 billion (-19.08%) QoQ, with the contraction mainly from on-chain borrowing. Notably, this marks the first time since Q3 2023 that the size of CeFi outstanding loans has surpassed that of DeFi lending applications.
Note: There is a risk of double-counting between total CeFi book size and DeFi borrowing statistics. Some CeFi institutions use DeFi protocols to lend to off-chain clients. For example, a CeFi institution might stake idle BTC, borrow USDC on-chain, and then lend that USDC to an off-chain borrower. This loan would be counted both in DeFi outstanding loans and as a loan to a client on the institution's financial statements. Due to a lack of disclosure and on-chain identity tagging, it is difficult to filter out such double-counting.
The QoQ contraction in DeFi lending was greater than in CeFi, erasing DeFi's former size advantage. By the end of Q2 2026, the market share of DeFi lending applications fell to 47.05%, down 555 bps QoQ; at the end of Q1, this share was 52.6%.
In the third major segment—the crypto-collateralized portion of CDP stablecoins—the size declined by $1.09 billion (-7.86%) QoQ. There is also a risk of double-counting here, as some CeFi institutions may obtain funds by minting CDP stablecoins and then lend them to off-chain clients.
Overall crypto-asset collateralized lending contracted by $11.33 billion (-16.78%) in Q2 to $56.16 billion, down 40.13% from the Q3 2025 peak of $78.69 billion.
Market share breakdown at the end of Q2 2026:
- DeFi Lending Applications: 36.37% (down 544 bps QoQ)
- CeFi Lending Platforms: 40.93% (up 324 bps QoQ)
- Crypto-Collateralized Portion of CDP Stablecoins: 22.7% (up 220 bps QoQ)
Grouping DeFi lending and CDP stablecoins together as the on-chain lending sector, their combined market share is 59.07%, down 324 bps QoQ.
Additional Perspectives on Decentralized Finance Lending
Outstanding loans in DeFi lending applications have been shrinking since hitting an all-time high of $47.13 billion on September 19, 2025; as of July 21, 2026, the size was $21.94 billion, a decline of $25.19 billion or 53.45% from the peak.
Since the end of Q1 2026, the pace of the drawdown in DeFi borrowing has intensified, though recent signs suggest a slight moderation.
Stablecoins
Using a 7-day moving average, the weighted average borrowing rate for stablecoins increased by 27 bps between March 31 and June 30; after the quarter ended, the rate continued to climb to 3.88%.
This metric is a weighted average, calculated based on the outstanding loan size, combining the borrowing cost from lending protocols and the minting fees for CDP stablecoins.
The chart below breaks down the two types of costs: borrowing stablecoins via lending protocols and minting CDP stablecoins using crypto assets as collateral. The interest rate trends for both are highly correlated, but the CDP minting rate is less volatile as it is set manually and periodically, not adjusting in real-time with the market. For over 21 months, both rates have been supported from below by the US Federal Funds Rate.
The USDC OTC benchmark rate ranged between 4.25% and 5% in Q2; it remained at 4.25% at quarter-end, a level that persisted until August 3.
The USDT OTC lending rate also fluctuated within the 4.25%-5% range.
Bitcoin
The chart shows the weighted borrowing rates for Wrapped Bitcoin (WBTC) across lending applications on multiple public blockchains. On-chain WBTC is primarily used as collateral, with weak borrowing demand, resulting in consistently low borrowing costs. Unlike stablecoins, on-chain BTC lending rates are very stable, with users borrowing and repaying infrequently. In Q2, the on-chain BTC lending rate fluctuated between 0.44% and 0.5%.
The historical spread between on-chain and off-chain (OTC) BTC lending rates persisted this quarter. OTC market demand for BTC lending stems from two sources: 1) demand for shorting Bitcoin, and 2) using BTC as collateral to borrow stablecoins or fiat. Demand driven by shorting is not common in the on-chain lending market, creating this on-chain vs. OTC cost differential.
The BTC OTC rate remained unchanged at 1% this quarter.
ETH and stETH
The chart below shows the weighted borrowing rates for ETH and stETH (staked Ethereum generated by the Lido protocol) across lending protocols and public blockchains. Historically, ETH borrowing costs have been higher than stETH's due to stronger borrowing demand for ETH.
Users frequently borrow ETH to execute looping leverage strategies: staking stETH (a receipt for staking ETH via Lido), borrowing ETH, and gaining leveraged exposure to Ethereum's staking annual percentage yield (APY). Under normal market conditions, ETH borrowing costs typically fluctuate within about 50 bps of the Ethereum staking APY. Once the borrowing cost exceeds the staking yield, this strategy becomes unprofitable, making it difficult for the borrowing APY to sustainably exceed the staking yield.
Similar to WBTC, stETH is mostly used as collateral, so the cost to borrow stETH is typically low.
Users can borrow ETH at very low or even negative net interest rates by using yield-bearing Liquid Staking Tokens (LSTs) or Liquid Restaking Tokens (LRTs) as collateral. This gives rise to the classic looping strategy: repeatedly depositing LSTs/LRTs as collateral, borrowing unstaked ETH, staking that ETH to obtain new LSTs/LRTs, and borrowing more ETH to amplify staking yield exposure. This strategy works when the ETH borrowing cost is lower than the staking APY obtainable from LSTs/LRTs. Except for a few special periods, this strategy has mostly been viable.
ETH OTC Lending Rate
Similar to BTC: the cost to borrow ETH in on-chain applications is generally lower than in the OTC market. Two main reasons:
- Borrowing demand from short sellers in the off-chain market is not common on-chain;
- The Ethereum staking yield acts as a floor for OTC lending rates: asset providers are unwilling to lend assets off-chain at rates below the staking yield. In contrast, in the on-chain market, the staking yield often acts as a ceiling for ETH lending rates.
Analysis of the Aave Lending Book
The following provides a filtered, in-depth analysis of the book for the Aave V3 core instance (currently the largest on-chain lending market). This analysis applies three filtering rules:
- Minimum Debt Threshold ($100): Positions below this threshold are excluded from aggregated statistics, filtering out insignificant small positions but biasing results towards larger loans.
- Health Factor (HF) Reporting Cap (HF ≤ 50): Positions with a health factor greater than 50 in the snapshot are excluded from core statistics; these over-collateralized positions are typically small and less meaningful for risk analysis. For thresholds within this range, debt-weighted average HF and percentile statistics only include positions where 1 ≤ HF ≤ 50; positions with HF < 1 are excluded from HF statistics, while positions with HF = 1 are included.
A higher health factor indicates a safer position; a health factor below 1 indicates the position has triggered liquidation conditions. Health Factor Formula = (Total Collateral Value × Weighted Average Liquidation Threshold) ÷ Total Borrowed Amount.
- Debt-to-Equity Ratio (D/E): Only positions where collateral value > debt (positive net equity) are included in D/E distribution and debt-weighted average D/E calculations. This rule is independent of the $100 debt threshold: even if a loan exceeds $100, it is excluded from D/E statistics if it lacks positive net equity.
Based on a snapshot from August 7, 2026, after filtering, there are 19,073 valid outstanding loans. "Efficiency Mode (e-mode)" loans constitute only 8.91% of the total positions but account for nearly 50% of the outstanding debt, similar to regular loans. E-mode is characterized by high price correlation between the borrowed asset and the collateral asset (e.g., ETH collateral borrowing WETH). In Galaxy's previous statistics from April 22, the e-mode debt share was closer to 60:40; the decline in share is due to a decrease in e-mode outstanding debt.
The table below shows the debt-weighted risk metrics for the filtered book, divided into three groups: all positions, e-mode, and regular mode. E-mode borrowers are highly leveraged overall: debt-weighted Loan-to-Value (LTV) is ~90%, debt-weighted Health Factor is only ~1.06, and debt-weighted D/E is ~10.7; meaning a slight shock to collateral prices could push many positions into stress territory.
Regular mode loans have a much thicker safety cushion: debt-weighted LTV is ~49%, Health Factor is ~1.79, and D/E is ~1.07; used to hedge risks unrelated to price movements between collateral and borrowed assets, e.g., using cbBTC as collateral to borrow USDC.
Debt-weighted D/E calculation formula (only for positions where Ci > Di, i.e., collateral > debt): D/E = Σi (Di × (Di ÷ (Ci − Di))) ÷ Σi Di, where Di represents single-position debt and Ci represents single-position collateral value.
Next, we analyze the USD value share of various collateral assets in the Aave V3 core market. ETH-related collateral dominates: WETH accounts for ~24%, weETH (Etherfi restaked ETH wrapper) accounts for 16%, wstETH (Lido stETH wrapper) accounts for 14%, with these three together constituting 54.6% of all usable collateral value; WBTC accounts for ~14%. A handful of assets thus underpin the majority of the collateral value in the book, with the remainder consisting of stablecoins and other yield-bearing tokens.
Looking at the composition of borrowed assets: WETH accounts for just over 37% of total liabilities, a typical outcome of the prevalence of ETH-related collateral looping leverage strategies. Stablecoin borrowing is also substantial: USDT accounts for ~28% and USDC ~22%, together comprising about half of total borrowing, with other tokens holding smaller shares.
Compared to the previous statistics, WETH's share of outstanding liabilities has significantly decreased from 51.1%, consistent with the earlier noted contraction in e-mode debt.
E-mode Sub-Perspective
E-mode collateral is highly concentrated in ETH staking/restaking wrapper tokens: weETH alone accounts for 42% of this category's collateral; combined with rsETH and wstETH, these three assets account for 66.2% of e-mode collateral value. This indicates that e-mode risk is not diversified collateral but essentially a concentrated bet on Ethereum's staking fundamentals.
On the borrowing side, e-mode debt is overwhelmingly denominated in WETH, with WETH alone accounting for 73% of e-mode debt, fully aligning with the user behavior pattern of using ETH-related collateral to loop borrow ETH. Stablecoins still hold a certain share, with USDT, USDe, and USDC together accounting for over ten percent of e-mode borrowing volume.
The table below details e-mode positions for each collateral asset, including debt-weighted risk metrics; for sub-samples where collateral is ≥99% concentrated in a single asset, the implied loop leverage count is also calculated. This is essentially a ranking of "which assets are most heavily leveraged through loops," different from a regular market cap ranking table. Liquid staking and restaking ETH tokens rank at the top, with high debt-weighted LTV, D/E generally in the high single to low double digits, and Health Factors barely above 1, corresponding to numerous tightly looped ETH leverage positions.
Implied loop count formula for a single position: N_i = ln((1 − (D/E)_i(1 − Li)) / Li) / ln(Li), where Li is the position's LTV; applied only to sub-samples where a single asset constitutes ≥99% of collateral. The output is the debt-weighted average Ni; positions where L is not in (0,1), the logarithmic parameter is not positive, or the result is not a finite number are excluded. Numerical output is only provided for liquid staked ETH, liquid restaked ETH, yield-bearing stablecoins, and Pendle PT tokens.
Corporate Debt Strategies
Galaxy Research currently tracks $16.1 billion in outstanding debt used by corporations to directly purchase or supplement digital asset treasury strategies. Due to limitations in Bloomberg's ability to track Strategy Inc.'s preferred stock, there is a time misalignment in the time series for the incremental circulation of the company's STRC shares, but the total debt figure accurately reflects actual liabilities.
Following Strategy's $15 billion debt buyback in May, outstanding debt for Digital Asset Treasury (DAT) enterprises decreased by $1.5 billion this quarter.
The table below shows the actual interest payments DAT issuers must make on their debt each quarter. Note that Strategy's STRC dividends require board resolution and must be paid from legally available funds; unpaid dividends accumulate and must be settled before distributions can be made to junior securities. Therefore, STRC dividend timing is irregular, and amounts are unevenly distributed.
Including DAT corporate debt, total industry-wide crypto-related outstanding debt declined 15.08% QoQ. Following the all-time high in Q3 2025, total crypto-related debt from all on-chain and off-chain channels fell to $73.2 billion at the end of Q2, marking the third consecutive quarter of decline.
Futures Market
Futures open interest (OI), including perpetual contracts, declined 3.08% QoQ to $103.2 billion; OI resumed its upward trend in July, reaching approximately $114 billion by month-end.
Important Note: Open interest size does not equal the absolute total leverage. Some open positions are delta-hedged with spot longs; OI alone cannot directly indicate the market's overall leverage ratio.
BTC futures OI fluctuated between $44 billion and $62 billion in Q2; starting the quarter at $48.04 billion, it fell to $45.04 billion (-6.24%) by June 30. After the quarter ended, it rebounded to around $48 billion by early August.
ETH futures OI declined more sharply than BTC's in Q2: starting the quarter at $29.84 billion, it was $21.99 billion on June 30, a drop of 26.31%; it rebounded to $25.74 billion after quarter-end.
At the end of Q2, combined BTC and ETH open interest stood at $67.07 billion, accounting for 65% of the total futures market.
Conclusion
We believe that following the significant decline in futures markets on October 10, 2025, Q2 2026 further confirms that crypto market leverage is being gradually digested. The lending market is descending in a stair-step fashion, not an elevator crash, with three consecutive quarters of mild contraction, rather than replaying the cliff-like plunge seen in a single quarter during the 2022 bear market.
Corporate treasury debt and futures open interest similarly reflect a controlled pullback, not a forced deleveraging. Early July data already suggests that futures OI and DeFi borrowing sizes may be nearing a bottoming range.
If this trend continues, the market possesses greater resilience to withstand further contraction, potentially avoiding the chain liquidations and counterparty domino failures seen in the last cycle. For now, deleveraging continues to progress step by step.
