Bridged USDe on MegaETH Fell About Three Quarters in a Day and Never Left Its Peg. The Loop That Drained MegaUSD Unwound One Level Up the Stack, at Par. Here Is What the Per-Chain Data Shows.

Earlier on July 10, 2026 we published a teardown of MegaUSD, whose supply on the MegaETH network fell by more than half in a day while its price never left a dollar. Sitting underneath that story is a larger number on the same chain that moved the same way and has drawn even less attention. The amount of Ethena's USDe bridged onto MegaETH fell from about 391 million dollars a week ago, to about 294 million a day ago, to about 72 million now. That is a drop of roughly 75 percent in the last 24 hours alone, and about 82 percent over the week. Across the entire move USDe traded within a few hundredths of a percent of a dollar. We read the current figure two independent ways before writing this, off DefiLlama's live per-chain series and straight off the USDe contract on MegaETH mainnet, and the two agree to within a fraction of a percent.

The shape is the same one we described in MegaUSD, one level up the stack. A dollar token whose price holds while its float leaves a chain has not depegged. Its holders on that chain have redeemed or bridged out at par. USDe did not break this week. It left. That distinction is the whole reason to read the per-chain supply rather than the price headline, because this week the two numbers tell opposite stories: the price says nothing happened, and the supply says a third of a billion dollars of float walked off one chain.

This is a direct companion to that MegaUSD teardown, and to our earlier walk through Ethena's USDe and the broader case for reserves you can verify. The method is the same across all three: read the chain, separate what the data proves from what it only suggests, and show the free check that would have let you watch this happen in real time.

What actually happened, and to which number

MegaETH is an Ethereum layer 2 built by MegaLabs, live on mainnet since February 2026, chain id 4326. USDe is Ethena's synthetic dollar, the largest of its kind, with about 4 billion dollars outstanding across every chain. It is not native to MegaETH. It is bridged there from Ethereum, and the amount sitting on MegaETH is a position that grows when capital bridges in and shrinks when it bridges out. That is the number that moved.

Per DefiLlama's per-chain series, USDe on MegaETH was about 391 million dollars on July 3 and held near that level for most of the week, easing to about 358 million by July 8. Then it fell in two steps: to about 294 million at the July 9 snapshot, and to about 72 million now. The reason it collapsed rather than drifted is the mechanism we named in the MegaUSD piece. MegaETH hosted a heavily incentivized loop, described in detail by analysts beforehand: deposit USDe, borrow rate-subsidized MegaUSD against it, swap the MegaUSD back into USDe, and repeat, for a leveraged yield reported around 25 percent at the top. A loop like that inflates the supply of both tokens on the chain, USDe as the base and MegaUSD as the borrowed leg, and the same commentators warned it would unwind quickly once the incentive compressed. When it did, both legs left at par: MegaUSD was redeemed, which is the story we told earlier, and the USDe that had been bridged in to feed the loop bridged back out, which is this one.

Measured across both Ethena-stack tokens on MegaETH, the footprint went from about 622 million dollars a week ago, roughly 391 million in USDe plus 231 million in MegaUSD, to about 189 million now, roughly 72 million in USDe plus 118 million in MegaUSD. That is about 433 million dollars, close to 70 percent of the combined position, leaving one chain in a week, with no price dislocation in either token at any point along the way.

Reading it off the chain, and the snapshot gap

Here is the figure we can stand behind without attribution, because we read it ourselves. The bridged USDe token on MegaETH is the contract at 0x5d3a1ff2b6bab83b63cd9ad0787074081a52ef34. Its symbol is USDe, it carries 18 decimals, and at MegaETH mainnet block 20,928,331 its totalSupply was about 71.89 million USDe. Anyone can reproduce that with one call against the public MegaETH RPC:

cast call 0x5d3a1ff2b6bab83b63cd9ad0787074081a52ef34 "totalSupply()(uint256)" --rpc-url https://mainnet.megaeth.com/rpc

Divide by ten to the eighteenth for the human figure. At USDe's price of about a dollar that is roughly 72 million dollars, and it matches DefiLlama's live per-chain field of about 71.85 million to within a fraction of a percent, the small difference being only that DefiLlama prices USDe at 0.9995 rather than exactly one.

One detail explains why almost no dashboard is showing this yet, and it is the same one that hid the MegaUSD move. DefiLlama's once-a-day historical chart still reads about 172 million dollars of USDe on MegaETH for July 10, because that chart snapshots once at 00:00 UTC and most of the float left after the snapshot. Its live per-chain field, and the contract itself, already read about 72 million. So a chart reading the daily series sees a manageable decline, while a chart reading the live contract sees more than half of what was left disappear in a few hours. That gap between the daily photograph and the live chain is exactly what a continuous read catches and a once-a-day aggregator does not.

One level up the stack, and the tell that it held at par

The MegaETH position is the sharpest part of a broader, gentler contraction in USDe. Per DefiLlama's daily series, total USDe outstanding fell about 158 million dollars in the single day from the July 9 to the July 10 snapshot, and the live figure is lower still, around 4.04 billion dollars as this is written, because the float kept leaving after that snapshot. Most of that decline is MegaETH: on the live reads, about 222 million dollars of USDe has left MegaETH since the July 9 level, and USDe's live network total sits below its own daily chart for the same reason the MegaETH number does, the float left after the daily photograph was taken. USDe's price over that window stayed near 0.9996 dollars, with a 24 hour range, per CoinGecko, of roughly a twentieth of a percent.

The cleanest tell that this was a leverage unwind and not a backing problem is a number that did not move. Ethena's own tokenized-Treasury dollar, USDtb, which Ethena issues and uses as a reserve asset for USDe, held flat at about 775 million dollars across the entire week, varying by less than a tenth of a percent while USDe shed hundreds of millions. If this contraction were a flight from the reserve, you would expect the reserve asset to move with it. It did not. That is consistent with redemptions and bridge-outs clearing at par against backing that stayed intact, which is the reassuring reading. It is not, and we will come back to this, proof that the backing is sound. It is evidence about where the stress was, not a solvency certificate.

What this is not

Three things this is not, stated plainly, because the words matter. It is not a depeg: USDe held a dollar on every venue we checked, on-chain and on CoinGecko, throughout. It is not an Ethena solvency event: the reserve rail held flat, no redemption was halted, and nothing we can find points to backing being impaired. And it is not an exploit: there is no hack, no drained contract, no attacker. It is capital that incentives brought onto one chain leaving that chain at face value once the incentive faded, which is the ordinary and healthy way a rented position is supposed to end.

What the data does not prove is the precise trigger, and we are not going to assert one. Whether an incentive was cut, a borrow rate flipped, or one large participant unwound first is not something the supply curve alone can tell you, and we would be guessing. We also cannot recompute USDe's backing from the MegaETH chain, because that backing lives off it, in Ethena's delta-neutral positions and reserves on other venues. A per-chain supply read is a real, cheap, early signal that a float is leaving. It is not a reserve audit, and anyone who tells you a supply chart proves a synthetic dollar is solvent is overclaiming in the other direction.

Where Kerne stands

The honest comparison here is structural, not a matter of size, and we have to say the size part first so the structural part is credible. Kerne is tiny. There are about 1,115 dollars of our kUSD outstanding, something like six orders of magnitude smaller than USDe, millions of times over. A 4 billion dollar protocol shrinking a few percent is still millions of times larger than everything we have built, and nothing in this piece is a claim that we are safer than Ethena at scale, because we operate at no scale at all.

What this week rewards is a property, not a size, and it is the property we build on. The float that left MegaETH was there because it was paid to be, and it went home the moment the payment stopped. The kUSD in circulation is backed one to one by USDC in an on-chain peg module, about 1,116 dollars of USDC against about 1,115 dollars of kUSD, a ratio you can read straight off Base and that we also publish, signed every hour, at our proof of reserves endpoint, with our own key and no attestor in the path. That backing does not walk off when an incentive compresses, because it is not an incentive. It is the collateral. The through-cycle yield we model on the staked version sits in an 8 to 9.4 percent band, and for context Ethena's own API puts the current staked USDe yield near 3.75 percent, but the number that matters this week is not which line is higher. It is that one dollar is redeemable against reserves you can recompute, and the other was rented deposits that were always free to leave. We are pre-audit, our first external review by Hexens begins in mid July and is not yet done, the delta-neutral leg of our design is currently flat rather than actively hedged at this size, and we disclose all of that rather than dress it up. None of this is an offer or advice. It is a description of what is on-chain and what is not.

What a holder should do, on any dollar

Not as an accusation, but as a checklist you can run on anything you hold, us included. First, read the price and the supply as two separate numbers, because this week they disagreed: a dollar token can hold its peg perfectly while most of its float leaves a chain, and only one of those two facts shows up in a price chart. Second, if you hold a bridged token, read its supply on the chain you actually hold it on, not just the global figure, because a position can look stable in aggregate while it is evaporating where you stand. Third, ask where the backing lives and whether you can recompute it, or only read someone's statement of it. We built a free tool to run these reads on any token or vault, at kerne.fi/verify, and we would rather you point it at us than take our word for any of the above.

Sources and as-of times. Current USDe on MegaETH: our own read of totalSupply on the MegaETH mainnet contract 0x5d3a1ff2b6bab83b63cd9ad0787074081a52ef34 at block 20,928,331, about 71.89 million USDe, July 10, 2026 around 23:20 UTC, cross-checked against DefiLlama's live per-chain field of about 71.85 million dollars. Per-chain and total USDe supply history, the roughly 391 and 294 million dollar prior levels, the 158 million dollar single-day total decline, and the daily versus live snapshot gap: DefiLlama stablecoin id 146. MegaUSD figures: DefiLlama stablecoin id 342 and our own read of the MegaUSD contract, about 117.6 million dollars now. USDtb flat at about 775 million dollars: DefiLlama stablecoin id 221. USDe price and 24 hour range, about 0.9996 dollars: CoinGecko, July 10, 2026. Staked USDe yield near 3.75 percent: Ethena's public yield API, as of July 8, 2026. The incentivized USDe and MegaUSD loop is a publicly described mechanism reported by analysts; we offer it as the most probable cause of the supply move together with the on-chain signature, and not as a confirmed trigger. Kerne figures: our own proof of reserves endpoints, July 10, 2026. Figures are point in time, on-chain supply is often revised, and USDe, USDtb and MegaUSD are not affiliated with Kerne. Recompute before relying on any number here.