DeFi Yield Update — September 28–October 4, 2026

1 Yield dashboard

2 Week-over-week comparison

Read the change column with a tolerance of roughly ten basis points, and this week every core row clears inside it — the first such week in our archive. Even the two rows that move most, Tron and Solana stablecoin lending, move less than last week's smallest exception. The absence of a signal is the signal: utilisation neither built on last week's lending demand nor gave it back.

Arbitrum easing to 5.20% is a two-basis-point move that leaves its spread at 1.75pp — the widest among tracked chains in every one of the ten weeks our archive covers.

3 Spread over the risk-free reference

Judged as spreads, the ordering is intact: Arbitrum pays 1.75pp, Base 1.17pp, Solana staking 1.46pp — a staking strategy out-earning most credit strategies in the sample, with a different risk content entirely — and Ethereum lending 0.76pp.

The row that should still give an allocator pause is ETH liquid staking at 2.19%, paying -1.26pp against the reference in the lending sense. That is not a flaw — staking is not a lending strategy, and its return includes token exposure rather than a credit premium. It does mean that choosing it on yield alone misreads what you are buying.

4 Lending rates by market

The spread between venues inside a single chain remains wider than the spread between chains, which is the practical argument for reading pool-level rates rather than chain averages. A composite will always look like something nobody can actually deposit into.

5 Liquid staking yields

Lido remains the dominant route by size, and its rate sits close to the sample average. Where providers differ by more than about twenty basis points, the gap is usually attributable to how much of the yield comes from MEV rather than from protocol issuance — which is also the component most likely to move without warning, as the week's MetaMask Staking incident outside this dataset reminded everyone: an operator-level fee-recipient redirect cost validators their tip stream while their stake sat untouched.

6 Stablecoin yields

Two things determine these levels, and only one is durable. The durable part is underlying real-world rate exposure from tokenized Treasuries. The cyclical part is perp funding, which feeds several of the higher-quoting products and can reverse in a day without anything visibly changing in the pool — and which the Ethena incentives change taking effect on 30 September puts a date against.

7 Playbook: matching strategy to mandate

8 Formulas used

9 What we would watch next week

• Whether the tokenized T-bill reference makes it four rises in a row; at 3.45% it does the compression work on its own whenever lending stalls.
• Whether sUSDe-linked rates flinch in the first full week after Ethena's ENA incentive stop — the cleanest test of a subsidised yield we will get all quarter.
• Whether Ethereum's premium holds near 0.76pp for a third week, which would make last week's re-widening look structural rather than event-driven.
• Whether Tron's dip to 3.52% is noise, by seeing it reverse or extend on a pool set thin enough to move on one venue's reporting.

10 In brief

This was a noise week, and the table proves it. Every core aggregate moved four basis points or less: Ethereum stablecoin lending rose from 4.17% to 4.21%, the tokenized T-bill reference climbed to 3.45% — its third consecutive weekly rise — and the blended stablecoin rate moved to 4.14%. The premium over the risk-free rate held at 0.76pp, where last week's re-widening left it. After two weeks of large moves in opposite directions, the market paused.

The smaller pool sets carried the week's biggest moves, and they point in both directions: Solana stablecoin lending rose about 16 basis points to 3.95% while Tron's fell about 14 to 3.52%. Both sit inside the sampling noise those thin pool sets always carry — which pools reported, how weighting shifted as TVL moved. The structure did not change; the noise did.

The one structural change worth naming came from outside the data: Ethena announced on 26 September that it will stop distributing ENA incentives that subsidise sUSDe yields, with the change taking effect on 30 September. That removes a subsidy from a slice of the stablecoin yield market, and the next few editions of this report will show whether the underlying basis trade carries the rate on its own.