PROTOCOL RESEARCH

Capital with a price

Plasma's savings vault took $2.7 billion in a single day at around 20% APY. It holds $32 million now, token incentives have gone to zero, and the remaining yield is close to 3.8%. That last number is the answer to what the 20% was made of.

When the vault launched in September 2025, capital arrived faster than almost anything in DeFi's history: $1.3 billion in the first hour, $2.7 billion within a day. Aave's deployment on the same chain took $5.9 billion in forty-eight hours and peaked near $6.6 billion by mid-October. The headline yield on the savings vault was roughly 20%.

A year later the vault holds $32.29 million — a little over one percent of what arrived on the first day. The easy reading is a story about capital leaving. The more useful one is about what that capital was being paid for, because it was visible on day one and is now measurable.

What the 20% was made of

The yield was never just yield. It was a stack: the underlying return on the deposited asset, additional return from looping, and XPL token incentives layered on top.

Maple's own reporting on its Plasma syrupUSDT vault makes the composition unusually legible. The vault realised roughly 12% annualised over its first two months, drawn from syrupUSDT's underlying yield, looping, and XPL rewards. The forward-looking figure was around 16% net — of which roughly four percentage points came from XPL rewards scheduled over the following thirty days.

Read that again. A quarter of the advertised return had an expiry date printed on it, disclosed in the fine print of a performance update rather than on the number people were actually looking at.

That distinction is the whole piece. Once yield strategies are stacked, the figure on the front end can describe something quite different from the economic engine underneath it.

What sits underneath

The stack on Plasma runs roughly like this, and each layer is worth naming because each adds a different dependency.

• Ethena supplies the base return on synthetic dollars, generated by running a delta-neutral basis trade — collecting funding from leveraged longs on perpetual futures. That pays while longs are paying. It stops, and eventually reverses, when funding turns negative.
• Aave supplies lending yield, and on Plasma it also enables the loop: deposit sUSDe, borrow dollars against it, deposit again. Each turn multiplies the base return and the liquidation risk together.
• Pendle lets the resulting positions be split into principal and yield, which adds tradeable exposure and a maturity date that the underlying strategy doesn't have.
• Veda packages all of it into a single vault with one number on the front.
• XPL incentives sit on top and are the only layer with a published end date.

In ordinary conditions these look like separate sources of return. In a sharp sell-off they are not. Funding flips negative, which hurts the Ethena leg. The same move pushes looped positions toward liquidation on Aave. Redemptions arrive at the vault while the strategies underneath are the hardest to exit. The correlation between the layers goes to one at exactly the moment the diversification was supposed to matter.

$160 for every dollar

Aave's retrospective on the deployment reports that the first eight weeks delivered around $160 of TVL for every $1 of incentives. It is presented as an efficiency figure, and by that measure it is a remarkable one.

Capital that arrives because the economics are temporarily subsidised is capital with a price. Remove the price, and watch what remains.

That is not a criticism of the number. It is the correct reading of it. A ratio that high does not indicate conviction; it indicates that the return on arriving was very high relative to the cost of leaving. Those are opposite properties, and only one of them survives the end of a programme.

The loop

The pattern connects to something I wrote about earlier this month: protocols that funded operations from their own token, and emptied out when the token stopped appreciating. This is the same structure one level up.

A protocol does not have to hold its own token for the token to become the engine. Here the depositor's return depends partly on XPL. XPL's value depends partly on the ecosystem retaining TVL. TVL depends on the incentives that XPL funds. Yield attracts capital, capital becomes TVL, TVL supports the narrative, the narrative supports the token, and the token funds the yield.

That loop runs in both directions, and the direction is set by something nobody in it controls.

The case for the other side

There is a real argument against reading this as a failure, and it deserves stating properly.

Bootstrapping a chain with incentives is a legitimate and well-understood strategy. Plasma used it to launch a working network with deep stablecoin liquidity from day one, integrations with Aave and Ethena, and a payments product that now ships a Visa card. None of that existed before, and none of it was going to be built by waiting for organic liquidity to show up. Paying for the cold start is a cost of entry, not a flaw — and the capital that left was doing what subsidised capital does, which the team would have known when they priced the programme.

The objection holds for the protocol. It does not hold for the depositor. Plasma got a network out of the exchange. The person who deposited at 20% got a yield that was partly a marketing budget, presented as a savings rate, next to the words beat traditional savings. Those are not the same transaction, and only one party could see both sides of it.

What to ask instead

When a dashboard shows a yield, the useful questions aren't about the size of the number.

• How much of it is protocol revenue — fees somebody actually paid?
• How much comes from leverage or looping, and what liquidation level does that imply?
• How much is token emission, and when does the programme end?
• If every layer of the stack had a bad week simultaneously, which of them could you exit?

Any product that can't answer the third question in a sentence is answering it by omission.

I set out to write this as an open question and the data closed it while I was working. Token incentives on Plasma are running at zero. The vault's fees over the last thirty days came to about $101,700, with no protocol cut, which annualises to roughly 3.8% on the current balance. That is the unsubsidised rate — what the stack produces when nobody is paying people to be there.

Roughly 3.8%, against a launch headline near 20%. The stack of Ethena, looping, Pendle and Aave is doing real work; it is doing about four percent of work. The rest was the marketing budget.

For completeness, the chain-level figures point the same way: Plasma generated $312 in fees in twenty-four hours against more than a billion dollars of stablecoins sitting on it. The liquidity stayed longer than the depositors did.

A 20% APY with a sustainable engine underneath is one product. A 20% APY where a meaningful share is temporary emissions is a different one wearing the same number. The figure on the dashboard is the surface. The composition is the spine.

Analysis, not investment advice. Vault TVL, fee and incentive figures checked against DefiLlama in September 2026; launch-period deposit and yield figures from contemporaneous reporting and protocol disclosures. The 3.8% figure annualises thirty days of fees against the current balance, which moved over the period. All of it changes continuously — verify before relying on it.