Ethena: how long it takes to exit sUSDe and what backs USDe

Perpetual futures—contracts with no expiry that track an asset's price—accounted for just 1.0% of USDe backing on July 3, 2026, with a yield of -0.1%: $39 million within a backing of nearly $4.5 billion. The mechanism that gave Ethena's synthetic dollar its name has been reduced to a residual component, replaced by lending, private credit, and AAA-rated securitized debt. On August 19, 2026, Ethena and institutional broker FalconX announced a $1 billion secured credit facility expanding this segment: it is the ceiling of a committed revolving line—as described in the contract reviewed fifteen days prior to the announcement—and the effectively drawn balance remains unpublished. What changes with this shift is the speed at which backing converts to cash, and that speed is set by a public formula: the sUSDe cooldown, the clock that decides how many days it takes to get your dollar back. This article applies that formula to the actual supply as of August 30, 2026, and lists the data series that track it live.

What did Ethena sign with FalconX and why is $1 billion not a disbursement?

The FalconX note was released on August 19, 2026, at 09:00 AM ET, and its headline refers to a warehouse financing facility—a wholesale financing line—to "expand institutional lending capacity." The body of the statement uses a different verb tense: it announces a $1 billion facility through a vehicle with Ethena "that will deploy capital from the assets backing USDe into overcollateralized institutional credit." Two paragraphs later, the same text claims the operation "marks one of the largest on-chain capital deployments into secured institutional credit to date."

Capacity is the ceiling of a line; deployment is the drawn balance. The distinction is resolved by the contract: the legal review that LlamaRisk—an independent risk analysis firm that audits the backing of several stablecoin issuers—published in the Ethena governance forum on August 4, 2026, fifteen days before the public announcement, describes the agreement as a revolving senior secured credit facility in which Ethena acts as the senior lender. Revolving means the vehicle draws and repays within the limit; the outstanding balance on any given day remains an unknown until Ethena publishes it.

The architecture, according to that same review: the borrower is FalconX International Lending Opportunities SPC on behalf of its segregated portfolio SP 1, a bankruptcy-remote vehicle (isolated from the group's bankruptcy) in the Cayman Islands. The vehicle uses the funds to purchase portfolios of crypto-secured institutional loans originated by two FalconX entities, and pledges those receivables—and all its other assets—in favor of Ethena. FalconX's triple role—originator, administrator, and collateral manager—and custody in qualified custodians are not detailed in that legal review, which only mentions the two originators: they appear in FalconX's own August 19 press release.

Relative size can be calculated. The Institutional Lending segment of the USDe backing was worth $310 million on July 3, 2026, representing 6.9% of the total. If the facility were to be fully drawn and everything else remained equal, that segment would rise to $1.31 billion, approximately 32% of the backing based on the supply of 4,079.9 million as of August 30, 2026. This is the arithmetic of the contracted ceiling against the last published snapshot: $1 billion committed on top of a $310 million segment.

What is backing USDe and what remains of the delta-neutral mechanism?

USDe was born as a delta-neutral dollar: spot purchase of an asset, simultaneous sale of the same amount in perpetuals, and the funding differential as a source of yield. That is the product we described in the May 2026 USDe analysis, when supply hovered around 4.6 billion following the rsETH—Kelp's restaked ETH—redemption window in April 2026. This piece updates it: the engine of that article has almost disappeared from the balance sheet.

The most recent official breakdown is the one from the June monthly governance report, published July 17, sourced from the transparency dashboard on July 3, 2026:

The stablecoins cell applies 35% to the backing of approximately 4,470 million; the report rounded it to "~2,000 million." INTX is the Coinbase international derivatives market.

The bottom line is what gives the product its name. The funding differential that sustained Ethena for two years was trading in the negative that day: maintaining the position cost money instead of paying it.

The complete series, with each figure tied to its source document, avoids the common dating error:

The "11% perpetuals" figure appearing in industry coverage stems from the March 12, 2026 governance proposal, and the official report confirms it as 11.4% at the close of that month. It is March data, not August, and between March and July, that proportion was divided by eleven. The curiosity lies in April: the percentage rose from 11.4% to 13.3% while the amount fell from 670 to 519 million, because liquid stablecoins were drained faster during the rsETH redemption window. The denominator sank faster than the numerator.

Two more dates complete the transition. On April 18, 2026, Ethena shifted its public taxonomy from two buckets (crypto base and liquid cash) to four, with DeFi lending and institutional lending as their own categories; the institutional segment was born that month with $10 million, or 0.3%. And on June 5 and 8, the first real-world collaterals other than Treasury bills were proposed: JAAA, the CLO strategy—collateralized loan obligations, corporate credit packages securitized by tranches—with a AAA rating from Janus Henderson tokenized via Centrifuge, and STAC, its equivalent from Securitize. We covered this step in the Janus Henderson agreement with Ethena; as of July 3, less than a month after those proposals, the category was already worth $501 million and represented 11.2% of the backing.

How long does it take to exit sUSDe and what formula sets the cooldown?

The cooldown for sUSDe—Ethena's staked USDe that receives yield—remains at one day when Tier 1 (assets liquidatable within 24 hours) covers at least 1.5 times the 99th percentile of daily redemptions, set at 4.1% of supply; as of August 30, 2026, that is $250.9 million. Since launch, exiting a sUSDe position required a fixed seven-day wait, and in March 2026 Ethena replaced it with this dynamic cooldown of 1, 3, 5, or 7 days recalculated daily. The proposal's reasoning: with 93% of backing in perpetuals, seven days described the balance sheet; with backing mostly in stablecoins, it was an arbitrary restriction.

The framework classifies backing cash into three tiers by liquidation speed. USDtb, which appears in two of them, is Ethena's second stablecoin, backed by short-term public debt. These are the proportions published in the May report, based on a liquid cash component that at the end of May 2026 was around $4 billion:

The 46% of backing placed in DeFi lending is split between Tier 2 and Tier 3 "according to real-time utilization conditions of each platform": withdrawing from Aave, Morpho, Kamino, or Jupiter is instantaneous as long as market utilization allows, and ceases to be so when that market tightens. The margin distribution and incentives driving that utilization are in the on-chain credit ranking by who keeps the margin.

The protocol's risk advisors calculated the 99th percentile of cumulative net redemptions for USDe's entire history: 4.1% of supply in one day, 9.2% in three days, and 12.0% in seven. Based on these benchmarks, the framework has three triggers, and only the first looks at the balance sheet:

1. If Tier 1 coverage—its balance divided by supply multiplied by 4.1%—falls below 1.5 times, the wait increases from one to three or five days depending on the drop.
2. If one-day requests exceed double the fourteen-day moving average and simultaneously three-day coverage falls below 1.5 times, the wait increases by one day.
3. If for two consecutive days outflows break the historical 95th percentile, it jumps directly to three days.

The timeframe is fixed at the moment the exit is initiated, so a tightening does not affect those already in the queue—nor does it protect them from the bottleneck that caused it.

The arithmetic for the August snapshot is straightforward. USDe supply was $4,079.9 million on August 30, 2026, according to DefiLlama, so the 99th percentile of daily exit equals $167.3 million and the minimum Tier 1 to sustain a one-day wait is $250.9 million. The buffer of stablecoins available for redemption shown on the dashboard on July 2—397 million in USDtb, 576 in PYUSD, 136 in USDC, and 128 in USDT—totaled about $1,237 million, but this buffer includes USDtb, and the framework places 99% of USDtb supply in Tier 2: once discounted, strict Tier 1 is around $844 million. Daily 99th percentile coverage therefore stands between 5.0 and 7.4 times, and the margin over the safety floor is between $593 and $986 million. The May report, with its own snapshot, estimated Tier 1 daily redemption capacity between 600 and 900 million and its coverage at 4-5 times.

The facility announced on August 19 has a $1 billion ceiling: it nearly doubles the Tier 1 margin in the strict reading ($593 million) and matches it in the broad reading ($986). The credit commitment and the buffer sustaining the exit window have become comparable magnitudes. Where it is funded from—the liquid buffer, rotation of the existing institutional segment, or new USDe issuance—will be revealed in the next monthly breakdown; the previous one was released on July 17 with data from July 2 and 3.

Is an overcollateralized loan as liquid as a perpetual?

Ethena holds two types of assets on the same balance sheet that are similar in solvency but differ in timeline, and the difference is best seen outside of crypto. Two savings entities have exactly the same money and the same capital. The first keeps it in demand deposits and three-month bills. The second has it placed in five-year mortgages, with conservative appraisals, first liens on the property, and zero historical delinquency. The second earns more and its portfolio is better secured. Both are solvent. Only one pays back on Tuesday.

The mapping to Ethena's balance sheet is direct and the composition table dates it: the 46% in DeFi lending and 6.9% in institutional lending as of July 3, 2026, are recoverable in two to five days according to the risk framework; the 1.0% in perpetuals, in minutes. A delta-neutral carry—the differential earned for holding both spot and perpetual—is closed in the same market where it was opened: the perpetual is repurchased, the spot is sold, cash comes in. A term loan to a trading firm, however healthy and overcollateralized, has its own schedule: it is recovered when it matures, when the borrower decides to prepay, or when collateral is liquidated, and none of those three things happen in minutes. Collateral can be worth 130 for every 100 lent and still not be in the account on the day redemptions come in a flood.

Overcollateralization protects against default, not illiquidity. These are two different risks with two different mitigants. Against default, first-rank security interests, qualified custody, and subordination of other vehicle creditors work. Against illiquidity, only one thing works: having assets that convert to cash within the window in which someone is asking for their money. A balance sheet with a 101.59% backing ratio can still take five days to convert its Tier 3 into cash.

Ethena's design turns this problem into a parameter. The dynamic cooldown explicitly recognizes that the user's exit speed must follow the backing's conversion speed: a slower balance sheet lengthens the timeframe in which the par redemption promise is fulfilled, without breaking it. This is why the risk framework organizes backing into 1, 2, and 5-day tranches.

What guarantees does the FalconX line with Ethena have and what is not published?

The legal review of the FalconX line identifies a first-rank security interest over all vehicle assets, a package of corporate separateness covenants typical of a special purpose entity, and the subordination of any other vehicle debt to Ethena's position. The information package is, in the reviewer's words, "the most robust in Ethena's lending portfolio": loan-by-loan data every business day, with the ability to verify collateral against underlying wallet addresses. The FalconX note adds that this collateral is held in qualified custodians. Guy Young, founder of Ethena Labs, described secured institutional lending as "one of the largest and most enduring sources of return in finance," and in terms of asset quality, the statement holds up.

What the structure does not publish is precisely the data that decides the liquidity question. The same legal review states that "specific contractual terms, commercial thresholds, portfolio parameters, and pricing clauses are confidential and not disclosed in this public summary." The loan term is among the reviewed sections, along with fund drawdown, reinvestment, and reporting, and was not made public: without an average term or early recovery conditions, an external observer cannot assign the segment to any of the three liquidation tiers of the risk framework: 1, 2, or 5 days.

Where does sUSDe yield come from if perpetuals no longer pay it?

The sUSDe yield comes from the composition table, line by line: 3.1% from DeFi lending on 46% of backing, 3.8% from liquid stablecoins on 35%, 5% from tokenized CLOs on 11.2%, between 4% and 7% from institutional lending on 6.9%, and -0.1% from the crypto base on the remaining 1.0%. The weighted average is around 3.7%, below what those who lock their coins earn: the yield of the entire backing is distributed only among those in sUSDe, a fraction of the supply.

The numbers from the protocol endpoint—last declared update August 26, 2026, retrieved on the 30th—fit this mix:

The last row explains the 11% yield figure still circulating—a historical average of 10.62% since launch, unrelated to the 11% perpetuals from March: it is dragged up by the exuberant funding quarters of 2024 and 2025. The current yield is 4.21% on a thirty-day average, and has been in the 4% zone for three months: 3.77% on June 1, 3.85% on July 1, 4.01% on August 19, 4.21% on August 26. The context of industry alternatives is in the stablecoin yield comparison.

This is the economic reason for the entire shift: when perpetual carry trades in the negative, keeping backing there destroys yield. Institutional credit at 4-7% and CLOs at 5% are the highest-paying lines in the table. The decision has financial logic: it replaces -0.1% with tranches paying between 4% and 7%, in exchange for assets recoverable in days rather than minutes.

What activates the ENA fee switch and how far is the first threshold?

Ethena protocol revenue has had a vote pending since August 27, 2026. The "ENA Fee Switch" proposal—the switch that activates protocol fee sharing—opened on Snapshot (ethenagovernance.eth) at 13:59 UTC that day and runs until September 2 at 13:59 UTC, scaling value capture with USDe supply: upon exceeding $7.5 billion, a 5% take of protocol revenue is activated for ENA buybacks, rising to 10% at 10 billion, 15% at 15 billion, 20% at 20 billion, and 25% from 25 billion, with buybacks tracked on the transparency dashboard. The two sources in the thread do not agree on the base for this take: the official table labels it on net revenue—the text allocates 95% of net revenue paid by business lines to the Foundation for buybacks—while the Blockworks Advisory analysis in the same thread applies it to gross revenue, as the table's own illustrative arithmetic does: 450 million × 5% = 22.5 million per year. As of August 30, the vote had accumulated 17.64 million ENA in favor, zero against, and 79 voters, with the 5 million quorum surpassed.

The distance connects it to everything above: with 4,079.9 million on August 30, the first threshold requires USDe to grow by 84%. It is a switch voted on but without current, hanging from the same supply that sets the Tier 1 buffer. That same week, the Foundation announced on August 27 the OTC buyback of locked ENA tokens from seed investors with more than 0.25% of supply who had sold since the October 2025 peak, and a one-time release of remaining investor tokens starting October 5, 2026. How it stands against the rest of the sector is in the fee switch and buyback comparison.

Where is USDe backing published and with how much delay?

Six public series provide live answers to what this article leaves open regarding USDe backing. None require privileged access; what changes between them is latency.

The order of that last row is the most useful finding: backing decisions are discussed and documented in the governance forum weeks before they exist as news, and those who read it are fifteen days ahead of those who read the press release.

Two additional signals. Liquidity in decentralized markets, the exit route for those who do not want to wait for the cooldown, is narrowing: USDe liquidity fell from 87.2 to 68.4 million dollars during June and sUSDe from 75.6 to 46.8 million, a minimal depth against a supply of over 4 billion. And the reserve fund, the buffer covering negative funding, stood at about 62 million in June, with a protocol backing ratio of 101.59%.

Would USDe withstand another redemption like October 2025?

USDe has already survived a redemption of that size. On October 11, 2025, amidst market tension, the protocol processed a daily outflow of $1,705 million, the largest in its history. That episode—with its secondary market price deviation—is what we analyzed in the May piece on Ethena's synthetic dollar. It was a real test, and it passed.

What is unknown is how the same test would behave against the August 2026 balance sheet, because the October 2025 balance sheet was almost the opposite of the current one: mostly delta-neutral positions that close in the same market where they open. The shift to lending, private credit, and securitized debt has not yet weathered a stress episode, and the risk framework itself acknowledges this by calibrating its tiers by conversion speed rather than asset quality. No forecast is possible here: there is no precedent. What does exist are the benchmarks to measure it when it arrives—the three-day 99th percentile is 9.2% of supply, $375 million; the seven-day is 12.0%, $490 million—and one unknown that conditions them: the loan term, declared confidential.

For the USDe holder, the practical consequence is concrete. USDe, born in 2024 as a delta-neutral dollar, is in August 2026 a diversified credit fund with a variable exit window governed by a public formula that anyone can recalculate. Monitoring the institutional lending line in the next monthly report and the current cooldown at the time of exit provides more information than any headline about the $1 billion.

Sources and links: FalconX via PRNewswire — official facility note (Aug-19-2026) · LlamaRisk — legal review of the FalconX credit agreement (Aug-4-2026) · Ethena — June 2026 governance report, with backing breakdown · Ethena — March and April 2026 governance report · Ethena — dynamic cooldown proposal for sUSDe and risk analysis (Mar-12-2026) · LlamaRisk — review criteria for institutional lending master agreements (Apr-14-2026) · DefiLlama — historical USDe supply (retrieved Aug-30-2026) · Ethena — ENA Fee Switch proposal, with milestone table and risk committee analysis (Aug-27-2026) · Snapshot — ENA Fee Switch vote, live count · Ethena — transparency dashboard and custodian attestations · Unchained — USDe reserves overhaul review · CoinDesk — FalconX agreement coverage