Aave v4 Risk Steward proposes cap increases and a flatter high-utilization WETH rate curve

What happened

LlamaRisk published a concrete Aave v4 Ethereum parameter recommendation through the official Risk Steward governance surface.

What changed

The proposal would expand USDT and USDG borrowing capacity, increase selected supply caps, reduce Core WETH slope2 from 14% to 8%, and close new deposits for matured PT-USDG-24SEP2026.

What did not change

The record does not establish execution. Aave v4 remains excluded by policy in the existing memo; no v3 parameter change or new approval is established.

Confirmed

• The proposal raises Core Main USDT add and draw caps from 28M and 20M to 40M and 30M.
• It raises the Core Ethena Ecosystem USDT draw cap from 1.5M to 3M and Core Main USDG draw cap from 50M to 60M.
• It raises Global Dollar USDG Maple USDG add and draw caps from 45M and 42.75M to 55M and 50M, and syrupUSDG add cap from 60M to 70M.
• It proposes Core WETH slope2 of 8%, retaining the stated 92% optimal utilization, 0% base rate, and 2.2% slope1.
• The proposed matured PT add cap is zero; the proposal states withdrawals would remain unaffected.
• The post describes implementation through the Risk Steward process as a next step.

Still open

• Execution transactions and resulting live Hub and Spoke parameters were not verified.
• The quoted liquidity and utilization observations are the steward's dated measurements, not independently reproduced current state.
• Post-change available exit liquidity and compliance with the memo's per-Spoke liquidity boundaries remain to be tested.

What it means for an advisor

• Monitor the aave-v4 diligence file for execution and reproduce the exact connected-Spoke caps and WETH curve before refreshing its configuration baseline.
• Retain the policy exclusion. Higher caps and a flatter high-utilization rate curve do not satisfy production-history or proposed-size withdrawal reopening criteria.
• Evaluate executed draw caps against matching-asset Hub liquidity rather than protocol-wide TVL.