The synthetic-dollar graveyard.

The table.

Circulating supply / market cap, peak to July 8, 2026. Every figure is recomputable from the public source noted at the bottom of this page; where independent trackers disagree, the range is shown. TVL and circulating supply are distinct metrics and are labeled as such.

Resolv USR: a points bleed, then one key.

USR's circulating market cap peaked at roughly $590M to $620M in early 2025 (DefiLlama recorded about $586.6M on February 21, 2025; CryptoRank about $621.5M). A larger figure sometimes quoted, about $684M, is Resolv's total protocol TVL, which includes the RLP overcollateralization layer, not USR's own supply; the two are different measurements and worth keeping separate. Through 2025 USR drifted down with the wider points-farming unwind, to roughly $340M by the end of December and about $103M by March 21, 2026.

Then the fast part. On March 22, 2026 an attacker who had compromised a single off-chain signing key, a privileged account that alone could authorize minting, deposited about $100,000 of USDC and minted roughly $80M of unbacked USR, orders of magnitude past the collateral, because the mint path had no on-chain check tying the amount to the deposit and no cap. USR fell from a dollar to about 2.5 cents within minutes, roughly $25M was extracted, and it never re-pegged; it trades near $0.15 today. USR is still tracked on DefiLlama, CoinGecko and CryptoRank rather than delisted, which is why it belongs in a graveyard and not just a history book. We walked the exact mint mechanism, and how Kerne's own mint path is gated against that specific class of failure, in the Resolv vs Kerne mint-path teardown; this entry is the size story, not the forensics.

Level lvlUSD: an orderly sunset, not a blowup.

Level is the honest exception in this table, and it is here precisely because it did not fail loudly. lvlUSD peaked at roughly $185M in circulating supply in May 2025 and sits near $515k today, about 99.7% below peak, but its dollar peg held near $0.999 the whole way down. There was no exploit and no depeg. The product is being wound down after the Level team joined Grove Labs, a protocol in the Sky and former MakerDAO ecosystem, and holders have been redeeming out in an orderly way.

We include a voluntary sunset alongside the collapses on purpose, because from a holder's balance-sheet view the line looks nearly identical: a dollar that was worth $185M is worth $515k, and the distinction between "the team moved on" and "the capital fled" only matters if you could see which one was happening in time. That is the recurring theme of this page. A dollar can go to near zero for a perfectly benign reason, and the only protection against being the last one holding it is being able to read what is actually happening rather than a headline about it.

Coinshift csUSDL and Neutrl NUSD: rented capital, on schedule.

These two are the clearest illustration of the pattern, because neither one broke. csUSDL, Coinshift's yield-bearing wrapper of USDL routed through Morpho, climbed past $100M and topped out at about $142M on July 1, 2025, at the height of its SHIFT airdrop points program. Its Season 1 snapshot closed at the end of July 2025, and the deposit base left almost exactly as fast as the incentive to stay did: about $100M by August 1, about $66M by September, about $6M by November, and roughly $53k today, about 99.96% below peak. It never lost its peg; it still trades slightly above a dollar because it accrues yield. The dollar did not fail. The reason to hold it did.

Neutrl's NUSD is the same story caught in the act. NUSD peaked at about $231M on February 16, 2026 and is near $68M today, roughly 70% below peak and down close to half in the last month alone. What makes it worth watching live is the timing: this is happening during Neutrl's own incentive Season 2, which runs into a planned token event, and whose published lock tiers advertise points multipliers of up to 400x and 700x per day. Capital is leaving anyway, near peg, while the incentives to stay are at their loudest. When 700x a day does not hold deposits, the deposits were never really yours; they were the points'. (The Season 2 multiplier figures are Neutrl's own published terms; the supply figures are recomputed from public on-chain data.)

What the data actually says killed them.

Line the four up and the delta-neutral mechanism is conspicuously absent from the causes of death. None of these dollars died because the basis trade stopped working. Two things recur instead.

1. Rented capital. Coinshift, Neutrl, and Resolv before its exploit all grew on points and airdrop incentives, and all shed that capital the moment the reward schedule turned. TVL that arrives for the points leaves for the points. It is not sticky, it was never underwriting the risk, and it exaggerates both the rise and the fall.
2. An over-powered control surface. Resolv's death had a second, sharper cause: a single off-chain key that could mint unbounded supply. That is not a delta-neutral problem; it is a trust-model problem, and it is the same one that ended several dollars we catalogue at /legible. The failure was always somewhere the holder could not see or check.

The mechanism these protocols share, holding an asset and shorting an equal amount of it to harvest funding, kept doing its job. What failed was the capital structure around it and, in one case, the keys above it.

The survivors, and the one trait they share.

Not every delta-neutral dollar is in the ground. Ethena's USDe is down roughly 70% from a peak above $14.8B in October 2025 to about $4.4B today, and it is very much alive: the largest synthetic dollar in the market, audited, with a real user base, and it kept operating through both a German regulator forcing a wind-down of its local entity and the largest asset manager in the world integrating it. Falcon's USDf sits near $1.26B, Avant's avUSD near $108M. These are not immune to funding cycles or drawdowns, and this page is not a claim that they are safe. But they did not evaporate.

The trait the survivors share, and the corpses lacked, is not size for its own sake. It is that the capital came for the product and can, to varying degrees, check what backs it. Ethena publishes collateral wallets and a weekly Chaos Labs Edge oracle attestation of delta-neutrality; we walk exactly what that does and does not let a holder verify in kUSD vs USDe and in the transparency scorecard. Durable capital plus a reserve someone can actually inspect is the combination that outlives a points program. Rented capital plus an unverifiable or over-privileged backing is the combination in this graveyard.

Where Kerne sits, honestly.

We said it at the top and it is worth repeating at the bottom: Kerne is smaller than everything in this table, roughly $1,100 in reserves across five holders, pre-audit, at Genesis scale. We have not survived anything, because we have barely been tested. So this is not "learn from them, we are the answer." It is narrower and more checkable than that.

We looked at the two causes of death and built against both. Against the rented-capital cause, Kerne is deliberately not running a mercenary points auction for TVL; the bet is on a durable anchor depositor, not a leaderboard, which is slower and is exactly why we are still at $1,100 rather than a rented $100M. Against the over-powered-key cause, the kUSD mint path is gated so no single off-chain key can mint supply: the mint privilege is held by on-chain contracts, the amount is derived from the contract's own USDC balance, and the admin is a 2-of-3 Safe, all checkable in a few calls against Base as laid out in Resolv vs Kerne.

And the honest boundary, because a graveyard is a bad place to hide one: kUSD's delta-neutral hedge runs on Hyperliquid, a single venue, and that leg is self-reported and signature-bound rather than independently re-derivable on-chain. It is our attestor-equivalent, an independent attestation of it is being scoped, and it is not yet live. Our reserve leg you can recompute; our hedge leg you currently take on a signature, and we say so everywhere, including in our open-gaps list at /legible. The difference we are betting on is not that Kerne cannot end up in a table like this. It is that if it starts to, you will be able to read it in the reserves before you read it in a headline.

Recompute the whole table yourself.

None of the numbers on this page are ours to assert. Circulating supply for every one of these dollars is public, and the death curves are just that supply over time. You do not need our word or a screenshot.

• For each dollar, DefiLlama's stablecoins API returns the full circulating-supply history: read the current value at stablecoins.llama.fi/stablecoins and the day-by-day series at stablecoins.llama.fi/stablecoincharts/all?stablecoin=ID (USR is id 197, Neutrl NUSD id 346, Coinshift csUSDL id 243). The peak is the maximum of the series; the collapse is the last value against it.
• Cross-check any current figure against a second independent tracker, CoinGecko or CryptoRank, before you rely on it. Where they disagree we show a range rather than a false-precision single number, which is the same discipline we apply to our own reserve reporting.
• To run the on-chain half of the same check on any dollar, including this cohort or your own holdings, paste an address into the free verify-anything tool, which reads live supply straight off the chain, and see the whole field ranked on verifiability in the scorecard.

Before you hold the next one.

The useful thing to do with a graveyard is to write down what you will check before you deposit into the next dollar, so the standard is set on a calm day. Is the capital here for the product or the points? Can you recompute the backing yourself, or only read an attestation? How many keys can mint, and what can each one touch? The free tools below run the on-chain half of that on anything you name.