A month ago AAVE was spoken of as the most bulletproof protocol in DeFi. Today the market prices it as if the protocol were on the brink. The whole question hides between those two pictures: did the machine break — or did fear create the discount?
What actually happened
Honestly, without smoothing it over. AAVE went through a rough stretch. After a recent exploit and the security problems tied to it, the protocol launched a recovery programme of roughly $300M to stabilise the platform.3 Bad debt was discussed, and part of the capital left Aave for more conservative protocols — users were cutting risk.
Add to that the legal saga of frozen ETH linked to a North Korean hack, and the departure of BGD Labs — the team that built Aave V3 — from the DAO amid governance conflicts; it was called the most serious loss of talent in the protocol's history.
The price reacted accordingly: from its prior levels, AAVE slid to around $95. The market is voting with its feet.
This is not background you can skip. It is the starting point of the thesis.
What did not break
And now — what the market's fear seems to have written off.
AAVE's money mechanism stayed in place. The DAO approved a standing token buyback funded directly from protocol revenue — around $50M a year on the original guidance.2 These are not inflationary rewards but real cash buying supply off the market. According to DefiLlama, the protocol generates roughly $95–100M of annual earnings on gross fees approaching $1B.1
The GHO stablecoin closes the loop: every GHO issued gives the DAO a fee spread, part of which flows to the treasury and into those same buybacks.4 More GHO in circulation — more revenue — more AAVE taken off the market.
And supply is tight: the token is capped at 16M, with about 15.38M already circulating, no future unlocks and no team vesting.5 What is on the market is essentially all there will ever be.
In other words: the protocol took a scar, but the machine that prints demand for its own token keeps running.
The valuation paradox
At a market cap of around $1.45B and annual earnings of ~$95–100M, AAVE trades at roughly 14–15× earnings. For a business that generates real cash, returns it to holders and dominates decentralised lending, that is the valuation of a company the market does not fully believe will survive.
The blunt question: is 14–15× the fair price of a broken business, or the price of fear around a business that will outlast the fear?
A Q3–Q4 price: a frame, not a prophecy
We do not name a "magic number." Instead, transparent arithmetic in which everyone sees our assumptions and can disagree. The target here is a function of two variables: whether earnings hold, and what multiple the market assigns.
An important correction to the "money machine" itself: the buyback the thesis leans on has already been trimmed. Even before the exploit, in February–March, the DAO cut its budget guidance from about $50M to $30M a year — because borrow fees had fallen roughly a quarter from the peak (January 2026: $7.95M versus $13.5M a year earlier).2 The machine has not stopped, but it spins slower than the headline picture suggests. That argues for caution on the multiple.
The second factor — the scar is real and measurable. TVL collapsed from roughly $26B to $14–15B — almost halving.1 Protocol revenue is a function of TVL, utilisation and rates, and the WETH rate compressed. So earnings in the coming quarters are not heading for a plateau but for a dip: we assume not $100M but a drop to ~$75–85M annualised at the trough, recovering to ~$90M by Q4 — but only if TVL starts to rebuild.
The third factor — and the decisive one: the ~$300M recovery programme is underway, but its outcome is not yet settled as of this writing. If it works and stops the outflow, the tail risk (the "death" scenario) leaves the balance sheet, and the arithmetic below holds. If it does not — the re-rating runs the other way.
At roughly 15.4M tokens in circulation:
| Scenario | Earnings | Multiple | Price per AAVE |
|---|---|---|---|
| Now | ~$95–100M | 14–15× | ~$95 |
| Cautious (trough) | ~$85M | 18× | ~$99 |
| Base by Q4 | ~$90M | 20–22× | ~$117–129 |
| Upside (TVL > $20B) | ~$95M | 25× | ~$154 |
Figures are illustrative and depend on earnings holding and the recovery programme working. If the exploit keeps squeezing TVL and revenue out, levels below current ones apply.
Base target: ~$100–130 by Q4, upside of about 5–35% from the current ~$95. The logic is a re-rating from a depressed 14–15× to 18–22× as the recovery physically completes and the fear of "death" fades. We deliberately do not bake in 25–40×: the revenue trajectory and the trimmed buyback do not yet justify it. This is a bet on a return to normal, not on euphoria — exactly what the whole thesis is about.
What flips us to the upper bound (closer to 25× and ~$150): TVL returns above $20B and earnings hold firmly above $90M. What flips it down is in the section "What would prove us wrong."
We believe the market is mistaking a scar for death.
The exploit and the outflow are real, and we are not downplaying them. But AAVE's money machine — buybacks from revenue, real yield near $100M, the GHO cycle and a hard-capped supply — remained intact. If the protocol stabilises after the recovery programme and TVL stops leaking, the discount fear has hung on the best franchise in decentralised lending will start to close.
We are betting on a re-rating recovery, not on a "memecoin 100×." This is a trade about a return to normal, not about euphoria.
What would prove us wrong
This is not a token ad. Here is what we are watching — and what would close the thesis:
Another exploit or a rise in bad debt, and the "scar, not death" argument collapses. Trust in a lending protocol does not recover after a second blow.
If TVL and revenue keep falling after the recovery programme, then capital does not believe in stabilisation — and the money machine will stall along with the fees.
The departure of BGD Labs is already a blow to expertise. If further governance fragmentation or a slowdown in V4 development follows, the foundation of the thesis weakens.
If at least two of these conditions turn against us, we will book the loss and publish the piece. As always.
Why this thesis carries no calendar horizon: it is event-driven. It closes on the conditions above — stabilisation after the recovery programme plays it out; the failures listed here invalidate it. It does not get to sit "Open" in silence: while it stays unresolved, every material development is logged as a dated entry in the Updates & Outcome block below.
This is analysis, not investment advice. Do your own research. The thesis is not changed after the fact.