sUSDe Market Consolidation and First Collateral Factor Reduction

[edo]

Summary

This proposal is the first governance action flowing from the RWG’s Complete Risk Refresh Assessment of USDe/sUSDe Collaterals on FiRM. It seeks to execute the first stage of the RWG’s phased sUSDe collateral transition proposed in the assessment. Three changes take effect simultaneously: the standalone sUSDe market is offboarded by reducing its supply ceiling to zero; the collateral factor on FiRM’s two FeedSwitch-protected sUSDe LP markets (DOLA/sUSDe LP and yv-sUSDe/DOLA) is reduced from 92% to 91.5%; and market ceilings are brought in line with current borrow utilization — DOLA/sUSDe LP to $50.5M and yv-sUSDe/DOLA to $5M. The existing 100% liquidation factor and 4% liquidation incentive is preserved on both LP markets.

Background

FiRM currently operates three sUSDe-denominated markets. The two LP markets — DOLA/sUSDe LP and yv-sUSDe/DOLA — are protected by FeedSwitch V2, which prices sUSDe at USDT-equivalent in its default configuration. A standalone sUSDe market with a $5M ceiling and zero active borrowers operates outside FeedSwitch coverage, pricing directly against the live Chainlink sUSDe-USD DEX-state feed.

The RWG’s Complete Risk Refresh Assessment documents five converging risk vectors across the sUSDe collateral stack. The structural finding central to this proposal is a 74% contraction in sUSDe DEX liquidity (ex-DOLA) since January 2026 — from $109M at FeedSwitch deployment to $28.34M as of April 20, 2026. This is the liquidity environment that underpins both the Chainlink sUSDe-USD feed and every FiRM liquidation path for sUSDe collateral. The 92% CF and the current market ceilings were calibrated against the January environment. They no longer reflect the conditions under which the protocol is operating.

The three changes in this proposal are addressed together because they belong together. The standalone market’s offboarding is a consolidation step — it removes a surface with zero borrowers and no FeedSwitch protection, focusing the remaining plan on the two markets where the FeedSwitch actually operates. Ceiling alignment with current borrow utilization closes the gap between structural ceiling and actual exposure, removing headroom that serves no function during a managed transition window. The CF reduction is the substantive risk adjustment — the first of two 0.5% steps that will bring the LP markets to 91% CF ahead of any feed configuration decision.

Why Now

The LP markets were parameterized at 92% CF against +$100M in independent exit liquidity. That pool depth is now roughly one-quarter of the environment the current parameterization was calibrated against. The Chainlink sUSDe-USD feed derives its price exclusively from on-chain DEX pools; feed behavior under stress is inseparable from the depth of the pools it sources from. Despite the recency of the October 2025 event, the current depth of $28M has no stress-tested performance record, and a comparable event would hit an environment with roughly 4× less absorbing capacity.

The FeedSwitch currently insulates LP market borrowers from real-time sUSDe price exposure. It was deployed as a temporary guardian-controlled buffer pending a trustless, PoR-driven automated fallback. That fallback has not materialized, and the current environment makes the interim posture progressively less defensible. Maintaining the FeedSwitch requires a named guardian to monitor, judge, and execute faster than a stress event deteriorates — a reactive model whose correctness is required exactly when the conditions for exercising judgment are most adverse. Even if exercised perfectly, guardian activation is still a reaction to an underlying problem already in motion and therefore cannot substitute for appropriate parameter sizing against the actual liquidity environment. Maintaining a 92% CF and ceilings unconstrained by utilization against a 74%-contracted liquidity picture is not consistent with the RWG’s collateral framework.

The RWG will conduct direct outreach to active sUSDe LP borrowers ahead of execution, consistent with prior parameter-change communications.

Proposed Changes

Market	Parameter	Current	Proposed
Standalone sUSDe	Supply Ceiling	$5,000,000	$0
DOLA/sUSDe LP	Collateral Factor	92%	91.5%
DOLA/sUSDe LP	Supply Ceiling	$80,000,000	$50,500,000
yv-sUSDe/DOLA	Collateral Factor	92%	91.5%
yv-sUSDe/DOLA	Supply Ceiling	$20,000,000	$5,000,000

Liquidation incentive preserved at 4% on both LP markets. FeedSwitch configuration unchanged. No changes to LF or daily borrow limits.

Combined sUSDe notional ceiling: $105M → $55.5M. Total ceiling reduction reflects consolidation to the two FeedSwitch-protected LP markets at utilization-appropriate levels.

Looking Ahead

If Stage 1 is approved via governance, Stage 2 — a second 0.5% CF reduction from 91.5% to 91% — is targeted approximately 30 days after this proposal executes, giving borrowers adequate time to adjust before the next step. During the Stage 1 observation window, the RWG continues structured data collection on Chainlink sUSDe-USD feed performance and sUSDe DEX TVL trajectory, and conducts engagement with Ethena on Season 6 status, sENA fee-switch activation timing, and forward guidance on the DEX-liquidity outlook.

If deemed necessary, Stage 3 — a data-driven ceiling reassessment and FeedSwitch configuration review — follows approximately 30 days after Stage 2, with a hard outer bound of June 26, 2026. It is a window, not a fixed execution date. The full rationale is set out in the Stage 3 proposal.

The RWG pre-commits to timeline acceleration under either of two conditions: sUSDe DEX TVL (ex-DOLA) sustained at $10M or below over a continuous one-week window, or confirmed migration of Aave’s sUSDe oracle across any of its Ethereum deployments from USDT-equivalent to live market pricing.

[cryptoharry]

I will vote against this proposal in its current form.

I agree with part of the direction here, particularly offboarding the standalone/naked sUSDe market. That market has no active borrowers, does not benefit from the FeedSwitch setup, and currently represents unused risk surface without much benefit to FiRM. Setting that ceiling to zero makes sense to me.

However, I disagree with reducing the ceilings for the DOLA/sUSDe and yv-sUSDe/DOLA markets.

These are among FiRM’s most utilized and profitable markets. It was not long ago that utilization was much closer to the existing ceilings, and I do not think the recent contraction should be treated as a reason to cap future growth at today’s lower levels. In my view, sUSDe remains one of the stronger collateral types available to FiRM, and I think we can safely support these markets at the current ceilings, potentially even above them, provided monitoring and FeedSwitch protections remain in place.

Ceilings are not only risk controls; they are also growth constraints. Reducing them this aggressively would limit one of FiRM’s strongest sources of productive DOLA demand.

On the CF reduction, I am more open to a measured decrease in principle. However, reducing the CF to 91.5% appears to push 0x86Dd1F01A8d94112b2427E1Ee13D9c5ea9A436d3 into liquidation. This user has close to $2m of debt on FiRM, and I do not think it makes sense to force a liquidation in this context. This is not an emergency offboarding, and the proposal itself frames this as a phased risk reduction. If the first step causes a large otherwise healthy borrower to be liquidated, then the step is too aggressive.

I would support a lower initial CF reduction that avoids immediate liquidation, with borrower outreach and further reassessment after the position has had time to adjust.

I also disagree with the framing around FeedSwitch. FeedSwitch is important infrastructure that protects users from unfair liquidations caused by temporary DEX dislocations, manipulation, or scam wicks. This is especially important given the market-risk profile of the live sUSDe/USD feed. I do not think FeedSwitch should be activated merely because there is some DEX stress. In my view, switching to live market pricing should be reserved for fundamental impairment scenarios, such as a loss of USDe backing, a serious collateral/custody issue, an exploit that mints unbacked USDe, or a breakdown in redemption reliability.

Short-term DEX stress is exactly the type of situation where we should not want users liquidated against a temporarily dislocated price. While FeedSwitch is using USDT-equivalent pricing, a temporary sUSDe DEX move does not mechanically create liquidation stress for FiRM, because FiRM is not using that DEX price for liquidations. That is the point of the design.

I also think the reduction in secondary sUSDe liquidity is being overweighted. It is worth monitoring, but it should be considered alongside Ethena’s move from a fixed 7-day cooldown to a dynamic 1–7 day cooldown. When the cooldown is at or near 1 day, the capital cost of arbitraging sUSDe back toward USDe is materially lower, which reduces the need for deep permanent secondary liquidity.

My preferred approach would be:

1. Offboard the standalone sUSDe market by setting its ceiling to zero.
2. Do not reduce the DOLA/sUSDe or yv-sUSDe/DOLA ceilings at this stage.
3. If reducing CF, set the first reduction at a level that does not immediately liquidate the large borrower above.
4. Keep FeedSwitch in place with the current default configuration.
5. Define clearer FeedSwitch activation criteria focused on fundamental impairment, not ordinary DEX volatility.

I appreciate the work RWG has put into this assessment, and I agree sUSDe should continue to be monitored closely. But I do not think the current data supports materially restricting FiRM’s strongest sUSDe LP markets. The standalone market offboarding makes sense, and a measured non-liquidating CF adjustment may make sense, but the ceiling reductions and the negative framing of FeedSwitch go too far in my view.

[karm]

Thanks for the considered response, Harry. The assessment and proposal cover substantial ground, and we welcome the opportunity to surface context that supports a more complete reading. Clarifications on the points raised below.

On 0x86Dd and the CF step. Fair point. We don’t want to force a liquidation on an otherwise-healthy borrower and will adjust the first CF reduction to a non-liquidating level alongside direct borrower outreach. We’d flag this as a recurring structural issue worth governance attention: as collateral risk profiles shift over time, FiRM has limited risk-reduction levers that don’t run into existing positions. The RWG weighs long-term parameter stability heavily when assessing new collateral for exactly this reason, since fixed-rate lending degrades quickly when parameters have to move sharply.

On ceilings and CF being correlated. The ceiling alignment and CF reduction belong together because increased position-opening or leverage during the transition window works directly against the risk reduction the CF step is meant to deliver. The current sUSDe demand environment depends heavily on the Ethena rewards system, and the forward path of seasonal rewards is unconfirmed. Liquidity has already shifted enough that we’d prefer not to introduce friction against the parameter correction during the observation window. Stage 3 is explicitly a data-driven ceiling reassessment. If the environment supports higher ceilings at that point, the proposal will reflect that.

On the FeedSwitch and the activation threshold for fundamental impairment. The live market feed sources only DEX liquidity by design. That’s the intentional path to solvency for a protocol that liquidates permissionlessly at block speed. Under FeedSwitch activation, that same DEX liquidity becomes the operative variable the moment the switch fires. If we agree FeedSwitch is reserved for critical scenarios, then DEX liquidity integrity matters most precisely at the moment it is under the highest volume.

If Ethena’s reserves face a custodial loss that impairs USDe backing, we should assume Ethena will use the emergency mechanisms designed for those scenarios, including gatekeeper-driven mint/redeem disable, role removal, and blacklist-and-redistribute. Losses exceeding the Reserve Fund surface as a USDe depeg in secondary markets. The cross-protocol pattern (Resolv, Elixir) under that condition is consistent: pause, snapshot, selective redemption favoring direct-channel and KYC-gated holders. This is congruent with the framing in the original FeedSwitch proposal: secondary dislocations resolve via arbitrage in normal conditions, not in the conditions when the protection is most needed.

Three considerations make FiRM’s position in such a scenario structurally different from Ethena’s own holders. There’s the standing question: FiRM may or may not be included in any recovery process, which is at Ethena’s discretion. There’s the composition question: the DOLA/sUSDe LP collateral is composition-sensitive, and the sUSDe share can shift materially under stress, particularly in the period following a snapshot block. And there’s the arbitrage question: the benefactor trade that supports sUSDe’s secondary price in normal periods can halt the moment impairment is suspected.

These considerations interact. They activate from the same trigger, in the same window, and they reinforce each other. Composition shifts because the arbitrage that would damp it has halted; a snapshot-based recovery could be keyed to a different composition than FiRM’s claim at redemption; standing discretion sits on top of both. The RWG operates monitoring infrastructure on each, and continues to develop it, but alerting and swift human intervention are not a substitute for parameter sizing under conditions that move this fast.

On the cooldown and the live-feed end-state. The dynamic 1–7 day cooldown is a real improvement, and it’s genuinely supportive of the live sUSDe-USD feed as the eventual destination. A tighter cooldown compresses the benefactor arbitrage cycle and damps routine sUSDe DEX deviations, which improves the DEX-state oracle’s behavior in normal conditions. The RWG’s Q1 oracle study (88 rounds, ~24-hour heartbeat, no hop exceeding 0.1%) and the October 2025 stress reference (deviation-triggered correction within ~23 minutes against ~$109M ex-DOLA depth) show the feed performed correctly in the conditions it was exposed to, and those conditions are structurally easier on the feed today.

What the cooldown does not address is the variable that drives FiRM’s tail risk. Stablecoin failure events unfold on a minutes-to-hours clock, the cooldown improvement is conditioned on the same benefactor arbitrage that pauses under suspected impairment, and DEX exits bypass the cooldown entirely. It governs Ethena’s redemption pipeline, not the channel FiRM liquidations route through.

On the broader posture. The structural concern beneath this proposal isn’t FiRM’s design. It’s the permanent installation of alert-based, non-market pricing on top of that design. FiRM liquidates permissionlessly at block speed against on-chain prices; that’s the mechanism keeping DOLA fully collateralized. Reserve-backed assets like sUSDe sit behind a redemption stack that is KYC-gated, discretionarily paused, and selectively honored under stress, and the FeedSwitch in its current configuration replaces market price discovery with a stable reference that requires human intervention to retire. The result is a workable mechanism that the RWG continues to steward and develop, but not one we recommend relying on as a permanent protection against fast-moving stress, particularly given FiRM’s existing bad debt and the absence of a junior tranche to absorb tail losses.

We appreciate the engagement on this as it sharpens the proposal. Whatever the vote returns, we’ll adapt our risk strategy accordingly and swiftly bring forward revised proposals reflecting that direction.