Stablecoins USDS and its predecessor DAI are widely regarded as the most popular decentralized stablecoins in the crypto market after USDT and USDC. Despite their popularity, both USDS and DAI operate on a complex infrastructure that may raise important questions—and concerns—for many users. GetBlock AML Research takes a closer look at the mechanisms behind USDS and DAI.

Key Takeaways

• The widely cited $11.1 billion figure represents the combined supply of two tokens: approximately $7.1 billion in USDS and $4.1 billion in DAI. Both stablecoins are backed by the Sky Vat accounting system, with total issuance across the ecosystem reaching roughly $13 billion.
• However, 57.5% of that supply—about $7.4 billion—is created through special credit facilities known as Allocators. These mechanisms are approved through Sky governance and are used to fund affiliated protocols within the Sky ecosystem. Rather than being backed by locked reserves, they are supported by lending activity conducted within the ecosystem itself.
• Approximately 84% of all issued USDS never leaves Sky’s own infrastructure. Around $5.9 billion is currently held in the sUSDS savings vault, while a substantial additional portion is locked in staking and rewards contracts. As a result, nearly all issued USDS remains inside the project’s ecosystem.
• Over the past 90 days, the network processed approximately $350.9 billion in transaction volume. However, most of this activity consisted of internal fund movements within the system. More than half of the volume—roughly $189.8 billion, or 54%—was directed to just four addresses controlled by Sky. During the same period, USDS transaction volume was 31 times larger than its circulating supply, yet the vast majority of that activity appears to have been generated by internal operations.
• Retail participation in the sUSDS savings product remains extremely limited. Analysis of current balances shows that Sky’s own protocols control approximately 32% of all assets held in the savings vault. Another 34% belongs to four large wallets, each holding more than $200 million. Meanwhile, retail users—the audience the product was presumably designed for—collectively hold only about $32 million across 2,824 wallets, representing less than 1% of total funds deposited in the system.

How USDS Is Backed

Most major stablecoins follow a relatively straightforward model. An issuer holds reserves and mints tokens against those reserves. For example, every dollar of USDC is backed by one dollar held in reserve. The reserves exist first; the stablecoin is issued afterward.

USDS departs from that model in a significant way.

The Reserve-Backed Portion

Part of the system operates using a traditional structure.

The Peg Stability Module (PSM) currently holds approximately $4.8 billion in USDC, which is invested in short-term U.S. Treasury securities. Users exchange USDC for USDS on a one-to-one basis, while the deposited USDC remains in reserve.

An additional $549 million is backed through collateral vaults where users lock assets such as ETH, wstETH, and WBTC in exchange for newly minted USDS.

Another $106 million is backed by vaults containing tokenized U.S. Treasuries and other real-world credit instruments.

All of these positions are managed by Sky’s central accounting contract, known as Vat. If collateral values fall below required thresholds, the system automatically liquidates the position.

The Allocator Model

The remaining 57.5% of supply—approximately $7.4 billion—is created differently.

Under the Allocator model, Sky governance votes to grant a debt ceiling to an affiliated protocol. Once approved, that protocol gains the ability to issue USDS up to the authorized limit.

Current allocations include:

• Spark: $4.1 billion
• Bloom: $2.7 billion
• OBEX: $608 million

At the level of the central Vat contract, these positions are not backed by specific collateral. Instead, the system records an obligation within an approved debt limit rather than a direct claim on locked assets.

Collateral does exist—but only at the next layer of the system.

Spark distributes newly created USDS through SparkLend, where borrowers post collateral such as ETH and wstETH with substantial overcollateralization. Bloom and OBEX deploy funds into short-term government securities and structured credit products.

This creates an important distinction.

When a borrower deposits ETH to obtain USDS through SparkLend, that ETH already serves as collateral for the borrower’s loan. At the same time, Spark issues USDS using a credit line that is indirectly supported by the same underlying lending portfolio.

In practice, the same pool of collateral contributes to two separate functions:

1. Securing the borrower’s loan.
2. Serving as the economic foundation supporting circulating USDS.

The key difference from a traditional collateralized model is not the existence of collateral itself, but how that collateral is controlled and enforced.

In the traditional model, collateral sits directly within the Vat system and can be automatically liquidated when necessary. Under the Allocator model, collateral exists outside the Vat framework. Losses can flow back into the Sky ecosystem, but there is no equivalent automatic liquidation mechanism protecting the system at the core accounting layer.

There is also a liquidity consideration. Assets such as government bonds and credit instruments cannot be sold as quickly as crypto assets like ETH. In a scenario involving large-scale redemptions, this could create additional liquidity pressure and make it more difficult to meet withdrawal demand immediately.

SparkLend’s Dominant Participants

Blockchain data provides a more detailed view of how the system operates.

Currently, 792 SparkLend borrowers have posted approximately $5.2 billion in collateral against just $586 million in outstanding debt, resulting in an average collateralization ratio of roughly 8.9x.

For comparison, most decentralized finance (DeFi) lending protocols typically operate with collateralization levels in the 1.5x to 2x range.

One address stands out in particular. It holds nearly $3 billion in collateral while borrowing only $71.6 million, resulting in a collateralization ratio of approximately 42-to-1.

Another major position belongs to Abraxas Capital Management, a London-based investment firm that reports managing approximately $3 billion in assets and specializes in ETH-backed stablecoin arbitrage strategies. This address contributes an additional $547 million in collateral.

Together, these two participants control more than two-thirds of all collateral deposited in SparkLend.

This structure suggests that SparkLend may function less as a traditional lending platform and more as a tool for managing large-scale capital reserves. Participants deposit substantial amounts of ETH, borrow relatively small quantities of USDS, and deploy capital elsewhere. These are not typical retail users.

At a collateralization ratio of 42-to-1, the liquidation mechanism becomes largely irrelevant in practice. ETH would need to lose roughly 97% of its value before such a position approached liquidation thresholds.

More importantly, a participant controlling assets of that size would likely be able to withdraw funds long before automatic liquidation became relevant.

Liquidation mechanisms are designed to protect a protocol from unintentional defaults. They do not protect against a scenario in which a major participant simply decides to exit the system.

There is another important consideration.

Spark’s $4.1 billion debt ceiling is based on existing collateral levels. However, only $586 million has actually been borrowed. In other words, only about 14% of the available capacity is currently being utilized.

The gap between the approved debt limit and actual lending activity amounts to roughly $3.5 billion. In theory, this represents additional USDS issuance capacity that could be activated beyond the platform’s current borrowing needs.

Whether the Allocator model represents a sophisticated architectural design or a hidden structural risk ultimately depends on how the system performs during periods of heavy withdrawals and how quickly governance can respond under stress.

The system has already survived the 2022 crypto market collapse, the USDC depeg in 2023, and several significant governance-related crises. However, the absence of a fully automated loss-absorption mechanism remains a fundamental characteristic of the design.

How USDS Is Actually Used

After issuance, most USDS rarely circulates outside Sky’s own ecosystem.

Approximately 84% of all circulating USDS—around $5.9 billion—is deposited in the sUSDS savings vault.

Another $831 million is held in the SPK rewards distribution contract, while an additional $207 million sits in a staking contract.

Taken together, Sky-controlled contracts hold nearly the entire USDS supply.

Transaction data from the past 90 days reinforces this picture.

Total transaction volume reached approximately $350.9 billion, a figure that may initially suggest widespread adoption and usage.

However, more than half of that volume remained within the ecosystem itself:

• Spark: $86.5 billion
• Sky Savings: $43.2 billion
• MakerDAO Governance: $36.0 billion
• Sky Treasury: $24.1 billion

Combined, these four destinations received approximately $189.8 billion, representing 54% of all transaction volume.

Over a 90-day period, USDS transfer volume exceeded its circulating supply by a factor of 31x, yet nearly all of that activity consisted of internal movements.

For comparison, Tether (USDT) is actively used by tens of thousands of wallets, exchanges, merchants, and payment providers worldwide.

DAI circulates across a much broader range of external DeFi protocols and user wallets. RLUSD also exhibits transaction patterns more consistent with real-world payment networks.

The $350.9 billion transaction figure for USDS tells a different story—one centered largely on the internal mechanics of a single ecosystem.

Who Is sUSDS Really Designed For?

At first glance, the sUSDS value proposition appears straightforward.

Users deposit USDS, earn yield through the Sky Savings Rate, currently around 3.65% annually, and retain the ability to withdraw funds at any time.

Those who have held funds since the product launched have earned approximately 9.77% cumulative yield over the past 20 months.

The product works as intended and has generated returns.

The more interesting question is: who is actually using it?

The current ownership structure of sUSDS provides some revealing insights:

The largest share of sUSDS is controlled by Sky’s own protocols.

Spark alone holds approximately $1.4 billion in sUSDS. Since launch, the protocol has processed roughly $10.7 billion in deposits across nearly 9,500 transactions.

The Sky Treasury provides an even more striking example. Around $7 billion has flowed through the treasury over time, while its current net balance stands at only $109 million.

This gap between cumulative flows and current balances helps explain the enormous volume of internal fund movements observed throughout the ecosystem.

The four largest wallets each hold more than $200 million in sUSDS. Together, they control approximately $1.7 billion, representing nearly one-third of the entire savings vault.

While blockchain data does not reveal the identities behind these addresses, the size of the positions strongly suggests institutional participants.

All remaining categories—including large investors, external protocols, and retail users—collectively account for roughly 34% of the system’s total assets.

That percentage represents the ecosystem’s actual external demand.

Retail users, despite being the audience for whom sUSDS is marketed as a savings product, collectively hold only $32 million, or less than 1% of all deposited funds.

In practice, sUSDS currently resembles a treasury and liquidity management tool for large institutional participants far more than a mass-market savings product.

What the Blockchain Reveals

The widely cited $11.1 billion figure combines two tokens backed by a single system.

More than half of that supply is issued through governance-approved credit facilities rather than traditional reserve-backed mechanisms.

At the same time, nearly all circulating USDS remains within Sky’s own infrastructure.

The savings vault, which is positioned as a product for a broad user base, is primarily controlled by the project’s own protocols and a relatively small number of large institutional participants.

This does not necessarily represent a flaw in the system.

However, these characteristics are important for anyone evaluating USDS as a financial instrument—whether banks, investment funds, corporations, or individual users.

Going forward, three questions deserve particular attention:

1. Will utilization of Spark’s $4.1 billion debt ceiling increase significantly?
2. Will the largest anonymous sUSDS holders begin moving funds in a coordinated manner?
3. Can retail participation grow beyond its current level of less than 1% as the product’s yield profile evolves?

The answers to these questions may provide a clearer picture of how resilient and decentralized the USDS ecosystem truly is over the long term.