Aave founder Stani Kulechov presented the changes as already underway, citing "low adoption." Each chain pulled in under $5K in quarterly protocol revenue — three of them under $1K. For oracle operators, this is a live autopsy of what happens when TVL thins past the operational break-even line.
The Decay Curve
The six-month deposit drops from the proposal:
- Sonic: -74%
- Scroll: -86%
- zkSync: -88%
- Metis: -79%
- Soneium: -95%
- Aptos available liquidity: -94%
Total exposure on the block: $98.1M supplied, $15.6M outstanding debt. 21 matured Pendle Principal Tokens — fixed-term products already past expiry — ride along in the cleanup.
Chain utilization dropped below the gas overhead of price updates. The feed economics went negative long before the proposal landed. LlamaRisk frames it bluntly: aggregate activity declined to a level where revenue does not cover the cost of supporting the deployment. The entire market winds down at once. No per-reserve triage.
The Revenue Floor
Hard numbers on income vs. operating cost:
- All six deployments: under $5K per quarter
- Metis, Soneium, Aptos: under $1K each
Price feeds, monitoring systems, and other operational costs consume those figures before any reserve-level revenue accrues. The cost of running aggregation workers, deviation checkers, and cross-chain message relays on these chains has exceeded the annualized yield. The honest framing is simpler than "low adoption": the data infrastructure bill had no unit economics.
The Wind-Down Mechanics
No cliff-shutdown. Aave is throttling the runway:
- New deposits, borrows, collateral switches: frozen
- Supply and borrow caps compressed to one token per market
- 5% base borrowing rate introduced
- 99% of borrower interest routes to Aave treasury
Suppliers earn less. Borrowers pay more. Both groups exit naturally. No liquidation cascade. No forced migration window. The price feeds simply go quiet. Liquidity stays locked until positions close organically.
What Oracle Operators Should Watch
If you route feeds across these chains, expect the following signal pattern:
- Heartbeat traffic degrades as reserve utilization scrolls toward zero
- RPC node count stays constant — your cost per active feed drifts upward
- Deviation thresholds stay static while underlying liquidity compresses — staleness check failures rise
- Cross-chain aggregation drift if Aave was a primary consumer on your routing topology
- Gas overhead per update remains pinned by chain complexity while revenue contribution drops to noise
The worst-case scenario: your feed goes stale during a liquidation event on a frozen market. Liquidators monitor the same pools. A two-second heartbeat miss on a chain carrying $15.6M in outstanding debt is the wrong metric to miss.
The April Kelp DAO hack forced this cleanup. The data infrastructure bill was already unsustainable — the hack just accelerated the formalization. Run the math on your lowest-revenue chains. The break-even threshold on price feed operations is now a public benchmark.