Aave is not pausing Ethena’s USDe. On 31 August 2026 at 19:25 UTC, LlamaRisk posted an Aave V3 Risk Steward batch that treats maxed supply caps as a plumbing problem: raise the cap where utilization is pinned at 100%, then make USDe borrowing 100 basis points more expensive so the same loop is not free. That is a usage print, not a distress print.
Ether (ETH) was about $2,465.55, up 2.02% over 24 hours, and Bitcoin (BTC) was about $78,547, up 1.13%, on the up&down CoinGecko snapshot fetched at 2026-09-01T00:03:57+00:00. Those prices set the tape. The Aave post sets the structure.
I care about the fee and the cap, not the roadmap. This batch is the fee and the cap.
What happened
LlamaRisk’s forum post, titled “Risk Stewards: Cap and IRM Changes on Aave V3 / 2026.08.31,” recommends a short list of parameter changes and says the next step is implementation through the Risk Steward process. It is not a token-holder vote. It is the delegated risk path Aave already uses for cap and interest-rate tweaks.
On Aave V3 Monad, PT-AUSD-8OCT2026 sits at 100.0% of a 40,000,000 supply cap. LlamaRisk wants that cap at 80,000,000. The reserve is collateral-only and lives in E-Mode category 5 (PT_Agora__Stablecoins) at 93.0% loan-to-value and a 95.0% liquidation threshold. The top eighteen suppliers all carry debt. Their health factors sit between 1.01 and 1.32, with a median of 1.02. LlamaRisk’s read is that this cluster is the steady state of a dollar-versus-dollar pair, not a crash in the making, because both legs are stablecoin-denominated.
Also on Monad, USDe is at 100.0% of a 120,000,000 supply cap, with outstanding supply listed at 120,001,531. The recommendation is 220,000,000. That reserve is collateral in E-Mode category 2 (USDe_sUSDe__Stablecoins) at 90.0% LTV and 92.0% LT. Eighteen of the top twenty suppliers have debt. Borrowing is in USDT0 and USDC against USDe.
On Aave V3 Plasma, USDe is at 99.8% of a 375,000,000 cap (374,072,211 outstanding). LlamaRisk wants 450,000,000. Concentration is high: the largest position is 73.0% of the reserve, and the top five hold 82.8%. Health factors on the top ten sit between 1.01 and 1.05, median 1.03. LlamaRisk still calls liquidation risk low because USDe is priced at par with USDT, so the health factor tracks the USDT peg rather than a directional coin.
The same post raises the USDe base variable borrow rate from 1.00% to 2.00% on five deployments: Core, Plasma, Monad, Mantle, and Avalanche. LlamaRisk’s table (same 31 August 2026 review) puts current borrow APRs at 3.85% on Core, 3.20% on Plasma, 2.33% on Monad, 2.09% on Mantle, and 9.13% on Avalanche, each +100 bps after the change. Optimal utilization is left unchanged, so this is a price change, not a quantity change, on the USDe borrow curve.
To absorb any USDe debt that migrates into other stables, the batch lifts optimal utilization from 92.00% to 94.00% on Core USDC, Core USDT, and Plasma USDT0. LlamaRisk quantifies about 116.63 million units of extra borrowable capacity below the kink, and it is honest about the other side of that move: withdrawal liquidity at the optimal point falls from 8.00% to 6.00% of supply.
WETH slope 1 is cut toward 2.20% on Core (from 2.35%), Arbitrum (from 2.50%), and Optimism (from 2.50%), aligned with a TokenLogic WETH rate note that LlamaRisk cites. Core WETH utilization is listed at 84.8% versus a 92.0% optimal. That is closer to the kink than a sleepy reserve.
Context
Aave remains a fee machine on Ethereum even while this cap work is happening on Monad and Plasma. DefiLlama’s Ethereum fee overview, pulled at 2026-09-01T00:14:50Z, shows Aave V3 at about $936,890 in 24-hour fees and about $6,461,962 over seven days. Aave V4 is a rounding error next to that ($15,191 in 24 hours on the same pull). The protocol TVL series on DefiLlama last printed about $18.09 billion at 2026-08-31T22:41:59+00:00, with about $15.35 billion of that on Ethereum (ex-borrowed). Show me the fee. The fee is still on V3.
Ethena USDe is not a toy float. DefiLlama’s stablecoin table, read at 2026-09-01T00:14:07Z, put circulating USDe at about $4.12 billion (peggedUSD 4,116,991,808). A companion stablecoin endpoint at 2026-09-01T00:14:50Z listed a USDe price of about $0.9997. Aave is where a large slice of that float gets reused as collateral. When Monad and Plasma caps pin at 100%, the venue is telling you the loop is full, not that the stablecoin vanished.
That is different from a bank-run story. In a run, people pull. Here, suppliers sit on health factors of 1.02 and wait for more cap. The risk is the usual loop risk: a peg wiggle, a concentrated wallet, a kink that leaves too little exit buffer. LlamaRisk names the Plasma USDT0 buffer as the line that warrants monitoring. That is the grown-up sentence in the post.
Traders who only watch Ethereum history and the ETH range will miss this, because the maxed caps are on Monad and Plasma. Ethereum Core still matters: USDe borrow APR on Core is the 3.85% to 4.85% step, and Core USDC/USDT get the 94% optimal. If you want a reminder of how we timestamp a tape versus a structure print, the desk’s market snapshot guide is the habit. For the ETH tape itself, the Ethereum news hub is the running log.
TVL is not the same as extra borrow capacity under an interest-rate kink. LlamaRisk’s 116.63 million figure is headroom, not new deposits. Confusing those two is how a cap raise gets sold as a TVL boom. The same split applies to liquidity: it is not the cap.
Our read
My stance: this is Aave adding room for a dollar loop that is already full, while charging more to borrow USDe. It is not a hidden unwind, and it is not a vote of no confidence in Ethena. The health-factor medians are tight because the design allows them to be tight. That can stay boring until a peg or a whale forces it not to be.
Falsifiable claim: by 2026-09-08 23:59 UTC, a later LlamaRisk or Aave primary should show the Aave V3 Monad USDe supply cap above 120,000,000 (the 220,000,000 room, or another figure that replaced 120,000,000). If that cap is still 120,000,000 on a primary, or LlamaRisk withdraws the 220,000,000 Monad USDe recommendation, this “stewards implement the room” read is wrong.
What to watch next
First, whether the Risk Steward path actually writes the Monad and Plasma caps. A recommendation is not a parameter. Until the cap moves, 100% utilization is still a closed door.
Second, Core USDC and USDT utilization versus the new 94% optimal. LlamaRisk’s point is that more USDe borrow cost may push demand into those reserves. If utilization jumps through the old 92% kink and sits on the steeper slope, the extra-capacity story gets a live test.
Third, Plasma USDT0. That reserve is already past its current kink in LlamaRisk’s table. The buffer cut is largest in relative terms where the absolute buffer is smallest. That is the line I would not ignore for a week.
Fourth, the ETH fee print on DefiLlama’s Ethereum overview and the ETH range on our snapshot. Aave V3 fees near $937k in a day say the core market is still paying rent. If that fee line drops hard while these side-chain caps stay pinned, the loop is migrating, not dying.
Cap raises are not a blessing. They are an admission that the last cap was the binding constraint. On 31 August 2026, LlamaRisk admitted that in public, with tables. I will believe the room when the cap number on Monad USDe is no longer 120,000,000.