What do you actually
get back?

1. Holding USDe: selling and redeeming are different

A sale needs another market participant, available liquidity and an acceptable price. Issuer redemption is a separate route with access rules. Ethena’s public terms distinguish ordinary Holding Users from eligible Mint Users. Owning the token does not automatically place you in the second group. USDe terms, updated August 2025.

For eligible users, the terms describe a reserve-based claim capped at one dollar, with conditions, rather than an unconditional promise to deliver a dollar immediately. The Mint User Agreement also addresses title to reserves and gives no unconditional redemption-timing warranty. A token balance is not the same thing as a bank deposit or direct ownership of a particular custody asset. Terms, including sections 2 and 14; Mint User Agreement, including sections 3 and 18.

Ordinary market exit: USDe → buyer or liquidity pool → the asset paid by that market.
Eligible issuer route: USDe → accepted redemption process → supported collateral delivery.

Those routes support one another only when the intermediaries can actually operate. Eligibility, fees, processing, available collateral and a buyer’s price matter. A discount can reflect difficult access without proving that the whole system is insolvent; assets valued above obligations do not prove that everyone can sell at par today. This distinction is the starting point for the rest of the report.

2. Staking: a second asset, then a second clock

Staking does not turn USDe into cash. In the selected sUSDe source, rewards are added to a vault and released into its accounting over time. The sUSDe balance represents shares: the amount of USDe associated with them depends on the conversion calculation, not a fixed one-token-for-one-token assumption. staking accounting; selected ERC-4626 conversion dependency.

The inspected cooldown path burns shares and moves USDe into a separate holding contract, the silo. The claim is then for that requested USDe amount, not continuing participation through the burned shares. Release depends on the recorded end time or the global cooldown being switched off. We have not read the live cooldown setting. StakedUSDeV2; USDeSilo.

sUSDe → request an unstake → USDe waits in the silo → release USDe → sell it or use an eligible redemption route.

A source-level detail that changes the wait: another request by the same direct account accumulates the amount and resets its end time. In a hypothetical 14-day setting, a request on day 0 followed by another on day 10 puts the combined request at day 24. This is an illustration of the inspected code—not a claim that today’s duration is 14 days. A wrapper can manage requests differently. request logic; a different wrapper’s request design.

Selling sUSDe instead is a market transaction with its own price and liquidity. It is not evidence that the underlying unstake is instant. The relevant comparison is the asset and proceeds available through each route, not just two buttons both labelled “exit.”

3. Nested tokens: maturity is not necessarily cash

Pendle separates a yield-bearing position into a principal token, PT, and a yield token, YT, using a standardized-yield wrapper, SY. The principal claim has a maturity and an accounting unit. “One unit at maturity” does not necessarily mean one sUSDe token, still less one dollar in a bank account. PT mechanics; SY mechanics.

Here the source-level detail is decisive: the inspected sUSDe-specific Pendle adapter permits sUSDe as its redemption output. We followed that override into the inherited output validation and transfer path. It does not make the final USDe-to-cash step disappear. This conclusion applies to the pinned adapter, not every Pendle market or newer chain implementation. sUSDe adapter; inherited vault wrapper; output checks and pause path.

A matching PT position → maturity/redemption rules → SY → sUSDe → unstake or sell → USDe → the final desired asset.

srUSDe adds yet another distinction. Strata’s senior token is not simply another spelling of sUSDe. Its documented structure adds a junior loss-absorbing layer and a strategy holding sUSDe. Junior support is finite, and output, accounting units, withdrawal rules and administrative controls still matter. This investigation read the design documents and a bounded Yearn assessment excerpt, not Strata’s implementation or current buffer. Strata risks; technical design; Yearn’s August 5, 2026 assessment.

A principal or senior label therefore identifies a position inside a structure. It is not enough to infer a risk-free dollar payment. A decline in a wrapper’s asset exchange rate can also affect a PT claim; that is separate from asserting that ordinary negative funding automatically reduces the sUSDe vault’s share conversion. Pendle’s negative-yield mechanics.

Borrowing adds a bill that does not disappear with the collateral

A loan against a token adds debt, a pricing rule and liquidation conditions. Liquidation means collateral can be taken and sold to repay debt when the loan’s conditions are breached. The accounting price used by a loan can differ from what a liquidator receives by selling. A deterministic PT maturity curve is a time discount, not by itself a current dollar price or sale quote. Its index and underlying price need compatible units. linear discount oracle; PT collateral and unwind path.

Illustrative arithmetic: $100 of your own equity supporting $500 of collateral and $400 of unchanged debt loses $25 if the collateral falls 5%. That is a 25% equity loss before interest, fees or liquidation. With an illustrative initial loan-to-value ratio—debt divided by collateral value—of 80% and a liquidation limit of 86%, the price-fall distance to the boundary is about 6.98%, not simply six percentage points. These are reproducible examples, not actual market settings or a suggested position.

In the selected Morpho core source, lender withdrawals require available liquidity. If a liquidated borrower exhausts its collateral with debt remaining, that bad debt reduces supply assets in the affected market. A vault holding positions in several markets adds further accounting and control choices; we have not audited those vaults or identified a current USDe market’s complete configuration. Morpho source, lines 180–440; liquidation mechanics.

4. Behind the token: who holds, owes and pays?

The familiar crypto hedge combines an asset holding with a short position, which gains when the referenced asset falls and loses when it rises, to offset price moves. That price offset does not eliminate funding costs, differences between the asset and hedge, margin needs, or the time required to collect a gain. Ethena also describes cash/stable assets, real-world assets, DeFi lending, institutional credit and non-crypto basis strategies. Their risks cannot all be reduced to one perpetual-futures trade. derivatives explanation; backing scope; institutional lending; real-world assets; non-crypto basis.

Off-exchange custody keeps custody and trading operations distinct. It does not make every settlement payment or withdrawal instant. The issuer describes different settlement arrangements by venue; Copper’s own service disclosures acknowledge interruptions or failed access. These are provider statements, not a legal or operational guarantee for the exact account backing USDe. settlement design; exchange-failure discussion; Copper disclosures.

Think of a simple timing mismatch. An asset worth $1,000 falls to $900 while a perfectly offsetting short produces a $100 receivable. On paper, the sum is still $1,000. If only the asset can be sold now, the short receivable does not immediately fund a $1,000 payout. This hypothetical ignores fees, basis and margin: it isolates why value and available money are different.

Lending can create another clock. March 2026 governance reviews of Maple/Anchorage lending discuss exposure limits and readily available assets. They distinguish the proposed Anchorage A1 legal entity from the US regulated bank. A brand name is not a substitute for the actual borrower, agreement and recall rights. The proposal and reviews do not prove funds were deployed. proposal, legal-entity discussion and risk reviews.

Issuer-reported DeFi lending also raises a question worth investigating: does backing depend on the same markets that use USDe-related collateral? If so, reduced lending liquidity and falling collateral could meet at a common exit. A shared protocol name is not enough to prove that loop. Exact positions, market identifiers and dates are missing here. Proposed Solana lending additionally requires its own chain and oracle checks. DeFi backing strategy; Solana proposal and selected review.

A conditional chain of stress—not a claim that it happened

Adverse carry or a borrower/venue problem → delayed usable backing → more holders seek a market exit → a discount or valuation mismatch → liquidations where loan rules transmit it → more selling into that route.

The chain can stop. Readily available assets can pay redemptions; effective settlement can turn a receivable into spendable funds; independent venues can offer another exit; collateral margins can absorb movement without liquidation. Their size, access and timing must be verified. An issuer reserve policy is a possible buffer, not insurance automatically available to each holder. reserve policy; funding-risk discussion.

A market discount alone does not establish a backing shortfall. Conversely, an apparently positive backing ratio does not demonstrate that a nested, borrowed or bridged position can leave without loss. Current exposure, executable depth and common dependencies are necessary before quantifying this stress path.

5. A backing number is only as good as its scope

The issuer’s November 2025 disclosure reports a snapshot at November 26, 2025, 23:59 UTC. Its listed components add to $7.295711 billion, versus the $7.297 billion reported aggregate. The $1.289 million difference is compatible with the figures’ displayed rounding precision. Using the aggregate and separately reported reserve produces approximately 101.00% of the stated USDe supply at the report’s dollar unit. historical publisher disclosure.

This is an arithmetic check, not an attestation. The embedded custody letters failed to load, so their signatories, scope and exclusions were not inspected. More recent attestation links were discovered but their underlying reports were not retrieved. These figures are neither a September 2026 balance sheet nor evidence of how much is accessible for immediate redemption. issuer attestation index.

What was checked—and what remains open

The investigation followed the historical Ethena token, V1 mint, staking, silo, administrator and one ERC-4626 dependency file; three selected Pendle adapter/base files; and the Morpho liquidation/withdrawal slice. It compared consequential historical audit findings with later source, read the public redemption terms and agreement, and examined backing, custody and lending documents. Exact source versions and limits are below.

The missing checks are material: current deployed-source matches and permissions; today’s cooldown, portfolios and reserves; underlying custody assurance; exact lending and oracle settings; executable market exits; and bridge implementations. The broader ecosystem is not fully inventoried or exposure-weighted. ENA, USDtb and cross-chain representations are not assumed interchangeable with USDe.

The useful first conclusion is about the position, not a safety label. Identify the asset you hold, what comes next, who must deliver it, how long that can take, and whether debt can force the timing. That is where a broad protocol story becomes a route a reader can actually understand.

Read the full inspection boundary · How evidence is classified

Sources and inspection limits

Retrieved September 11, 2026. A retrieval date is not an event date, a commit date or a measurement block. Issuer statements, provider documentation, commissioned audits and reviewer observations have different evidential roles.