In an era of declining returns, where does Sky's excess returns come from?

Written by: Vadym

Compiled by: Luffy, Foresight News

It's now an undeniable fact that on-chain yields have generally declined, and related discussions have been widespread. Lending rates are converging with Federal Reserve rates, and "safe" deposit rates average only around 3%, lower than US Treasury bonds and the Secured Overnight Financing Rate (SOFR). It's worth noting that at the time of writing, USDC and USDT yields on Aave are around 2%, Steakhouse and Gauntlet Prime offer around 3.2% (on-chain yields are more competitive), and sUSDe yields 3.5%.

Of the over $20 billion in vault assets on Ethereum and L2 networks, 58% of stablecoin total value locked (TVL) has an annualized yield below 3%, and 35% is between 3% and 5%. This makes Sky's USDS savings product, offering a yield of 3.75%-4%, a highly attractive safe haven. But the question remains: how can Sky offer above-average interest rates while possessing such a large TVL?

Data source: Galaxy Research, https://app.vaults.fyi/

Of course, directly comparing Sky's savings rates with Aave or Morpho's lending rates isn't entirely accurate. The latter's returns are entirely based on borrowing demand and capital utilization, while Sky's return mechanism is different:

• The interest rate (stability fee/stable savings rate, SSR) is set by the governance body.
• Sky can accept any collateral it approves for lending, including financing initiated from off-chain assets (however, this model is a double-edged sword, as lending rates within its ecosystem are less flexible and more expensive).

Nevertheless, all projects remain in competition on the funding supply side. Sky acts similarly to a central bank, participating in the vast majority of revenue streams by issuing unsecured credit lines to various entities. Therefore, exploring the nature of its revenue streams, the proportion of on-chain and real-world asset revenues, and the correlation and scalability of various revenue streams is of great value.

As Sky co-founder Rune himself said, "Currently, stablecoins worth over $300 billion are not generating any returns"... But how much of this huge potential demand can actually be absorbed?

On-chain revenue overview

Before delving into Sky's balance sheet and financial situation, let's briefly outline how on-chain revenue is generated and the overall revenue scale in 2025.

Simplified on-chain revenue sources

Loan interest

Like traditional finance, the money market is also at the core of DeFi, accounting for over 60% of the total TVL (TVL) of DeFi. In 2025, total lending interest revenue (including collateralized debt positions and uncollateralized/low-collateralized lending) was approximately $1.76 billion. So what is the core motivation for users to borrow? Currently, there are three main lending models: unified liquidity pools like Aave v3, segregated pools with risk managers like Morpho, and the central bank model of Sky and its related ecosystem. We will break them down one by one.

Top 15 borrowers on Ethereum Aave v3

Morpho's Top 15 Borrowers

There are three main types of scenarios driving lending demand:

• Revolving lending. This involves borrowing related assets and revolving them to maximize the returns on interest-bearing assets. Currently, in Ethereum Aave v3, approximately 39% of lending demand is used to amplify ETH staking rewards, with Fluid, EtherFi, and Lido being the main participants, historically exceeding 45%; sUSDe revolving lending demand accounts for approximately 11.6%. On Morpho, at least 27% of lending is directly used for USD revolving strategies within the protocol, earning returns on assets such as sUSDS, syrupUSDC, and sUSDe; approximately 5% is used for ETH-related strategies.
• Spread trading and leverage. Approximately 45% of Aave lending demand falls into this category: users borrow stablecoins using non-stable assets such as BTC and ETH as collateral, and then invest the funds in other channels to earn interest spreads, improving capital efficiency. For example, the largest borrower on Morpho (6.5%) borrows USDC and deposits it into Sky Savings products. Another major scenario is spot leveraged long positions (top borrowers on Aave, 0x54d25, xed0c6, 0x28a55, and 0x741aa, all use this strategy), or borrowing stablecoins for other purposes while maintaining collateral exposure. Currently, on Morpho, approximately 40% of interest income comes from lending using cbBTC as collateral (excluding the largest borrower), with a significant portion of these funds originating from Coinbase users.
• Other secondary needs include borrowing non-stable assets as collateral to short sell using stablecoins, or investment strategies based on other trading pairs.

As we can see, about half of the borrowing activities are essentially aimed at leveraging other sources of income. This raises two core questions: Where does this interest come from? And what alternatives are available once existing strategies are exhausted?

Public blockchain network staking rewards

The staking rewards for native assets on public blockchains are mainly divided into two categories: network issuance rewards and MEV rewards (priority fees + bribes).

Data sources: Dune, Helius, Blockworks

Ethereum's staking rewards primarily come from network inflation, with a maximum daily issuance of approximately 2,700 ETH, totaling around 1 million ETH by 2025. Solana's annual issuance is approximately 24-25 million SOL. The advantage of this type of reward is its high stability, but the principal price fluctuates significantly.

Of the staking rewards, 5%-20% (5%-30% on Solana) comes from on-chain activity (priority fees and MEV), a proportion that has been declining since Ethereum merged. This is because approximately half of MEV revenue comes from arbitrage and half from front-running; the latter has been significantly reduced due to the widespread adoption of anti-MEV tools such as OFA and solvers (currently, about 90% of transactions are private routes); the former was once monopolized by non-neutral searchers and block builders, but is now highly dependent on market conditions and highly competitive.

Funding rates

Ethena pioneered the on-chain perpetual contract funding fee revenue model, generating approximately $240 million in fee revenue in 2025 (90% of which came from funding fees), with total revenue from this model reaching approximately $300 million. The unique value of this type of revenue lies in:

• Introducing new revenue streams to DeFi through tokenization and composability;
• Profits are easy to capture;
• It has sustainability, even with significant market volatility (16% return in 2021, 0.6% in 2022, approximately 9% in 2023, and approximately 13% in 2024).

This makes it an ideal underlying asset for derivatives such as fixed-rate products and interest rate swaps, and related products have already been launched on Pendle.

While its scalability remains to be seen, BTC perpetual contract open interest is projected to be between $35 billion and $65 billion in 2025, while ETH perpetual contract open interest is estimated at around $20 billion to $40 billion. By early 2026, the total market open interest for perpetual contracts is expected to reach approximately $75 billion, with Ethena currently accounting for about 1.8%.

Transaction fees

Token swaps remain one of the core activities on the blockchain, providing infrastructure for traders and collecting transaction fees—a stable path to profitability. In 2025, automated market maker (AMM) liquidity providers generated approximately $4.2 billion in transaction fees, with 62% coming from Uniswap, Meteora, and Raydium.

However, capturing these benefits through structured products is not easy:

• Liquidity providers (LPs) often face losses: liquidity pools are vulnerable to malicious order flows, and centralized liquidity providers are particularly prone to losses, usually only professional players participate; the market acceptance of Uniswap LP management products is also not high.
• Limited composability: Limited application scenarios for LP positions as collateral.
• Profits are concentrated in a few assets: A large portion of transaction fees come from niche assets, making it difficult to adapt to products with ordinary returns. In 2025, approximately 25% of Ethereum's trading volume came from ETH-stablecoin trading pairs (currently around 60%), and 41% from USD trading pairs; on Solana, 50% of trading volume came from SOL-USD trading pairs, Meme coin accounted for 30%, and stablecoin trading pairs accounted for only 5% (currently a ratio of approximately 62:12:17).
• The battle between AMMs and PMMs: The scalability of AMMs remains controversial. Although most front-end trading volume is routed through the solver network, only about 11% is completed through Professional Market Makers (PMMs), with the remainder still handled by AMMs. Solana's dedicated AMMs had a 30% market share in 2025, which has now risen to 60%, dominating the SOL-USD trading pair, with a peak market share of 86%.

Nevertheless, there are still some trading strategy vaults on Gauntlet, and Sky, the core subject of this study, also holds some AMM liquidity provider positions.

With the increasing popularity of DeFi perpetual contracts, their market-making vaults (HLP, LLP, etc.) can be seen as an alternative to Uniswap LP management products. LP returns are moderate (approximately $130 million), with JLP contributing an additional $670 million. Risk managers are closely monitoring this area, whose risk characteristics can be summarized as follows: "Products such as HLP, LLP, Giga Vault, and OLP have experienced maximum drawdowns of 5%-9%, and have seen consecutive months of floating losses; daily returns conform to market-making characteristics, with most showing small profits, some breaking even, and occasional large losses; risk-adjusted returns are excellent, but the distribution of returns is uneven and path-dependent."

Risk transfer benefits

There are three ways to manage risk: retention, mitigation, and transfer. Currently, most participants primarily choose to retain risk and employ advanced mitigation tools. However, the channels for transferring volatility risk or protocol risk (technical, economic, and governance risks) remain very limited, and the associated premium gains are negligible. This may represent an untapped blue ocean market.

Decentralized options are not a new concept; the DeFi space has seen numerous attempts, including options AMMs, perpetual options, and options vaults (DOV), but most products have failed to stand the test of time. However, teams continue to develop in this area, and more tokenizable volatility-yielding products are expected to emerge in the future.

From a market opportunity perspective, CeFi options and perpetual contracts face fierce competition. CeFi options open interest is projected to be between $30 billion and $50 billion in 2025, with current on-chain options open interest at approximately $1.8 billion (primarily from Derive). Option premiums, like funding fees, are persistent, but capturing and packaging them is extremely difficult. It's worth noting that as risk managers' asset allocation becomes more sophisticated and competition intensifies, vaults may incorporate options into their portfolios to maintain a competitive edge.

Premium income from insurance underwriting remains minimal, with the vast majority coming from established projects like Nexus Mutual. In 2025, the platform generated over $5.5 million in revenue for insurance sellers, with risk exposure primarily concentrated in on-chain products such as Fasanara, Infinifi, and Dialectic. In the current low-yield environment, on-chain risk pricing is extremely complex, and the insurance sector is still in its early stages. Demand may grow as revenue sources diversify and protocols mature. Indirect mechanisms such as Aave's Umbrella insurance and extracting reserves from transaction fees can mitigate the insolvency risk of a single protocol, but they are difficult to adapt to structured products.

Real-world assets (RWA)

This article aims to analyze the sources of on-chain revenue; therefore, the introduction to RWA is only a rough estimate for market reference. According to data from rwa.xyz, RWA's total value has grown from approximately $5.6 billion in 2025 to the current $27 billion, with US Treasury bonds accounting for the largest share (approximately 41%) and private credit accounting for approximately 25%. By applying conventional yields to the average size of various asset classes in 2025 and including interest-bearing commodities and real estate income, RWA's total annual revenue is estimated to be approximately $600-900 million.

In summary, the total on-chain revenue in 2025 was approximately $8 billion, but the distribution of revenue sources was uneven, and some revenues, even if available, were difficult to capture reliably (such as AMM fees and volatility returns), while the potential for secure returns was even more limited. However, with the continued innovation in DeFi and the emergence of crypto-native opportunities, more on-chain fee and revenue capture mechanisms are expected to appear in the future.

Back to Sky

Let's get back to the main topic and explore how Sky integrates various revenue streams to form a coherent and stable strategy by analyzing its performance in 2025 and its current layout.

Sky's Financial Briefing for 2025-Early 2026. Data source: Dune, Sky Forum.

Simplified asset overview and projected return structure. Data source: Sky Official.

Currently, Sky derives approximately 55% of its revenue from its direct lending business (PSM, crypto vault, traditional RWA, and SKY token collateralized lending), with the remaining 45% coming from uncollateralized credit lines issued to its affiliated ecosystem.

PSM (Pegged Stable Module) remains the largest source of revenue, contributing approximately 44% of total income. In 2025, approximately 27% of PSM's total revenue will come from USDC deposits on Coinbase.

The returns from crypto vaults are moderate, with ETH and WSETH collateralized assets contributing the most.

Excluding PSM, direct exposure to traditional RWA has continued to decline, and revenue is currently mainly generated through the newly launched ecosystem partner Grove Finance. Of its cumulative revenue, 47% comes from Janus Henderson JAAA (cash-like instruments and mortgage-backed securities), 16.7% from Janus Henderson JTRSY (US short-term Treasury securities), and 15.2% from BlackRock BUIDL-I (cash-like instruments).

Obex is a new credit line partner, with approximately $600 million currently allocated to syrupUSD, and plans to become the core allocation platform for structured interest-bearing products in the future.

Spark's revenue and asset share have steadily increased since April 2025, surpassing the d3m lending model to become a core allocation channel, contributing 20%-65% of revenue. Spark acts as both an asset manager and a risk manager.

• Operate its own lending protocol Sparklend (an Aave v3 fork).
• Within the risk framework, flexibly allocate interest-earning opportunities that are higher than the borrowing cost (SSR + 0.3%).

So, where exactly does the revenue flowing into Spark come from?

Data source: Dune, Spark official website

Sparklend lending is the largest asset in Spark's liquidity layer, contributing 37% of total gross revenue (excluding reward distribution) in February 2026, largely consistent with the levels in the third and fourth quarters (33.4% and 37.8%, respectively). Why do users lend and borrow on Sparklend?

Sparklend's Top 15 Borrowers

Since only WSETH is an interest-bearing token that can be used as collateral, its core use is the classic revolving strategy, accounting for about 50% of the total lending volume and being the main source of reserve income. Major participants include institutions such as Ipor, Threehouse, Mellow, and Summerfi.

The majority of the remaining loans are in stablecoins. Only USDT's interest rate model is independent of SSR, based on capital utilization, which may explain its position as the second-largest lending asset (approximately 25%). Why are users willing to borrow other stablecoins at rates higher than the market average? The reasons are unclear. For example, the top lending addresses on Spark are associated with 7 Siblings, accounting for at least 15% of total lending (and holding hundreds of idle ETH). Their recent activities include leveraged long positions in ETH and SKY and participation in SKY staking; internal capital reuse, revenue sharing, and SPK mining are likely the primary motivations. The second-largest borrower continuously buys and stakes SKY using the time-weighted average price (TWAP).

More than half of the revenue comes from other income sources, with specific investments dynamically adjusted based on current market returns. For example, a significant portion of the revenue in the third and fourth quarters came from Morpho, Maple, and Ethena; currently, daily revenue comes from Maple (10.6%), Anchorage (11.2%), and PayPal deposit earnings (30%).

So how does Maple generate returns? Approximately 28% of the funds are used to issue collateralized loans to institutional borrowers at an annualized interest rate of 6%-9% (collateral being blue-chip assets such as BTC, XRP, and SOL). The remaining funds are then allocated to interest-bearing assets, with one-third flowing back into USDS savings products and the remainder invested in Aave, PYUSD, Superstate, and syrupUSD liquidity pools.

What about Anchorage? They offer BTC-secured loans to institutions at interest rates below 6.5%.

What about Morpho and Ethena? Their revenue logic has been analyzed in detail in the previous article.

Finally, a brief summary of Sky's financial situation: Total revenue in 2025 was $338 million, with $194 million paid out to savings users. The current savings pool of $6.7 billion in USDS has a payout rate of 3.75%, distributing approximately $688,000 in daily dividends and yielding approximately $1.17 million in daily returns. Based on the current projected revenue structure, Spark's revenue is roughly 50:50 between on-chain and off-chain sources, while approximately 70% of Sky's overall revenue comes from off-chain sources. What does this mean for the entire market?

Regardless of where the returns originate, a sustainable source of income has emerged on-chain, exhibiting lower dependence on market activity and leverage demand, and weaker correlation with traditional on-chain returns, providing a stable safe haven for individuals and protocols. For Sky, its diversified and flexible allocation strategy allows for rapid adjustments when on-chain risk appetite recovers. Furthermore, with USDS becoming the largest interest-bearing stablecoin in terms of real-time volume, combined with the behavior of leading borrowers, it can be seen that Sky indirectly helps other protocols maintain interest rate levels.

Some argue that traditional finance is swallowing DeFi, and indeed, many tokenized RWAs have access restrictions, making participation difficult for ordinary users. However, although yields are generated off-chain, they ultimately flow back to on-chain for distribution. This trend is crucial, bringing liquidity and diversified yields to DeFi, allowing permissionless protocols to benefit. Furthermore, this may drive the implementation and large-scale development of next-generation yield derivatives such as fixed-rate products, interest rate swaps, risk stratification, and structured products. Is a true DeFi renaissance imminent?