What can we do with wstETH to earn additional yield?

Scope: Ethereum mainnet · Position: $60M wstETH (26,259 wstETH) · Modelled deployment: $24M (40%) = 10,503 wstETH Date: 2026-08-03 · Block stamp: all on-chain reads at block 25,673,966 unless stated

Answer

There are exactly three things you can do with wstETH. Only one of them pays.

wstETH already earns 2.21% via the stEthPerToken() exchange rate, and that accrues wherever the token sits — wallet, Aave aToken, Morpho collateral, identically. Everything below is measured as incremental return over that 2.21% baseline.

Root cause of the poor economics: Ethereum staking yield has compressed to 2.21%, at or below what every liquid venue charges to borrow ETH (1.75%–2.58%). The leverage trade has no spread to multiply.

Recommendation: Path 0 (hold). If the mandate requires yield, Path B1 at 2.46%.

1. Market data

1.1 wstETH baseline

Aave’s implied wstETH/WETH ratio is 1.24080 against the on-chain exchange rate of 1.240774 — Aave prices wstETH off the exchange rate, not market price. Consequence: eMode liquidation is driven by Lido’s redemption rate, not by a stETH market depeg.

1.2 ETH borrow cost — every liquid mainnet venue

The two cheapest venues are the two that cannot take the size. No incentive campaigns are live — apps.aavechan.com/api/merit/aprs returns null for every mainnet key including ethereum-borrow-weth; DefiLlama shows zero reward APY on all these pools.

1.3 Stablecoin borrow cost (Aave v3 core, including our own $8.4M of demand)

Aave stablecoin reserves use slope1 = 4%, Uopt = 92%, slope2 = 100% — the rate climbs 12.5% APR per percentage point past the kink.

1.4 Stablecoin deployment venues

Every venue at or below the 3.71% borrow cost is safe; every venue above it is paid-for risk. sUSDS at 3.52%, Aave USDC supply at 3.24%, Steakhouse USDC at 3.27%, Gauntlet USDC Prime at 3.44% are all negative carry.

1.5 Exit liquidity

CoW Protocol quotes, wstETH→WETH, inclusive of fee:

A single large DEX order carries heavy price impact and cannot be routed above ~22,000 wstETH. Mitigate with TWAP: 500-wstETH clips quote at −2.6bp, so 24,000 wstETH costs ~$14,300 spread over ~1 day. CoW supports native TWAP orders (ComposableCoW); fills are gasless to the taker.

wstETH suits TWAP unusually well — the pair is a deterministic exchange rate (no timing risk) and is arbitrage-anchored to Lido redemption, so impact mean-reverts between clips rather than accumulating. Clip quotes are independent spot quotes, so spread execution over a day rather than an hour.

Lido withdrawal queue: 5,343 stETH unfinalized, last request 131,497 vs finalized 131,369 — effectively empty. MAX_STETH_WITHDRAWAL_AMOUNT = 1,000 stETH, so a $60M exit splits into ~33 requests. Zero-slippage path for size, a few days.

2. Option 1 — Lend wstETH: does not exist

Cause: lending rates are paid by borrowers, and nobody wants to borrow a yield-bearing asset — you would pay interest and be short an asset appreciating 2.21%/yr against ETH. Aave and Spark did not merely fail to attract borrowers; governance actively disabled wstETH borrowing, because a market with no borrow demand is a liquidation surface with no offsetting revenue.

Conclusion: wstETH yields 2.21% and only 2.21%, wherever it is held. Any additional yield must come from borrowing against it.

3. Path 0 — Hold

Baseline for every path below.

4. Path A — ETH-correlated loop

Mechanics: supply wstETH → enable Aave eMode 1 → borrow WETH → swap to wstETH → re-supply → repeat.

Formula: net APY = L × 2.21% − (L−1) × borrowAPY(debt), with the borrow rate re-solved on Aave’s IRM after adding our own debt.

IRM calibration: the model carries a +5.23bp offset so it reproduces the on-chain rate (2.1148% APR) at the observed 80.7431% utilisation. Aave v3.2+ computes utilisation off virtual balances; the raw curve understates the rate without this.

4.1 The dollar gain is a function of debt alone

gain$ = Debt × (2.21% − borrowAPY(Debt))

Borrowing more earns the spread on more notional while compressing that spread on all of it. Optimises at a fixed point independent of equity:

• Optimal debt: $54.9M (29,812 WETH)
• Maximum gain: ~$20,000/yr — hard ceiling at any equity size

No leverage at any size in the $18–30M range reaches +20bp. At 5x on $24M the gap is 4bp — the borrow rate would need to be 2.160% instead of 2.201%, or Lido 2.251% instead of 2.210%.

4.2 Position parameters at $24M

4.2.1 Round-trip friction

Entry is nearly free: buying 24,000 wstETH quotes at −0.8bp ($4,572), because aggregators arbitrage against Lido’s mint (getCurrentStakeLimit() = 150,000 ETH available, not paused). Exit costs ~$14,300 via TWAP or $0 via the Lido queue (§1.5). Both routes de-lever progressively — withdraw within LTV headroom, sell/redeem, repay, repeat — so no flash loan is needed.

Friction is not a reason to reject Path A.

4.3 Why the cheaper venues cannot be used

Only Aave v3 core has depth: 412,038 WETH ($758M) available, borrow-cap headroom 672,350 WETH, wstETH supply-cap headroom 107,370 wstETH (Path A uses 32%).

4.4 Borrow-leg rate risk

Aave WETH utilisation reached 98.48% at block 25,018,766 (borrow 4.239% APR). IRM parameters changed during the window — Uopt 94%→92%, slope1 2.00%→2.35%, slope2 3.00%→6.00% between blocks 25,119,566 and 25,321,166 — so the series is not strictly comparable. Under today’s curve, 98.48% utilisation implies ~7.21% APR, which would make Path A deeply loss-making.

Verdict: reject — but on the gain, not on friction:

1. ~$20,000/yr is the ceiling at any size (§4.1) — 8.3bp on $24M, and it does not scale with capital committed.
2. 3.29x leverage is required to earn it, with a $54.8M debt position to manage.
3. The borrow leg can reprice against you (§4.4) — WETH utilisation hit 98.48% within the quarter; under today’s curve that implies ~7.21% APR, which turns the position deeply loss-making.
4. Exit must be TWAPed or queued (§4.2.1) — a single atomic unwind at this size cannot be routed on-chain, so an emergency de-lever is expensive and slow exactly when speed matters.

Friction does not kill Path A — round-trip is ~$21k, paying back in 1.1 years. The trivial, hard-capped gain does.

5. Path B1 — All-USDT → Fluid fUSDT

Mechanics: supply wstETH (no eMode) → borrow USDT → deposit fUSDT. No swap, no basis risk.

Risks: ETH-price liquidation · Fluid utilisation (deposit not withdrawable if borrowed out) · Fluid smart contract.

5.1 Why not a higher-yielding USDT venue?

Steakhouse USDT (4.92%) and Smokehouse USDT (5.04%) both post higher headline rates than Fluid’s 4.41%, and both are native USDT — no swap, no basis risk. Rejected for two measured reasons.

(a) Our own deposit lowers the rate. These vaults lend into Morpho markets at 90–93% utilisation. Steakhouse USDT is 100% allocated to the wstETH/USDT 86% market (0xe7e9694b754c4d4f7e21faf7223f6fa71abaeb10296a4c43a54a7977149687d2): $138.0M supply, $127.0M borrowed, 92.0% utilisation. Since supply APY = borrow APY × utilisation × (1 − fee):

And it continues falling: Morpho’s AdaptiveCurveIRM targets 90%, so post-deposit utilisation sits below target and the borrow rate drifts down further over subsequent days. Fluid dilutes too, but $8.4M is 6.4% of fUSDT against a much larger shared liquidity layer.

(b) Exit capacity — the binding constraint. Pro-rata claim on idle market liquidity against an $8.4M ticket:

Supply caps are not the constraint — both vaults have ample headroom ($3.7B and $999B nominal). The constraint is that 90%+ utilisation means only ~8% of each market is idle at any moment.

Net economics after dilution: Steakhouse still beats B1 — roughly +33bp / $79,800 vs +24.5bp / $58,800. The extra $21k/yr is real. It buys a position that is 83% exitable on demand, in a vault 100% concentrated in wstETH — the same asset already posted as Aave collateral.

Smokehouse fails on every axis: 36% of a $23.3M vault, 0.21x exit coverage, 20% allocated to Pendle PT-reUSD (a maturity-dated token) and 9% to reUSD.

Decision: B1 keeps Fluid fUSDT. B1’s purpose in this study is the path where every risk is bounded and exit takes minutes; substituting a venue with 0.83x exit coverage converts it into a worse-executed B3. If the higher yield is wanted, B3 buys it more cleanly — +47.3bp with an explicit redemption queue and a secondary market, rather than +33bp with silent illiquidity.

6. Path B2 — USDT → swap → Fluid fUSDC

Steps 1–2 as B1. Then swap USDT→USDC (~1–3bp via CoW or Curve 3pool), deposit fUSDC at 4.83%.

Additional risk vs B1: USDT/USDC basis — you owe USDT while holding USDC. Normally <10bp; a 1% adverse move costs $84k, about a year of the edge.

7. Path B3 — USDT → swap → Maple syrupUSDC

Steps 1–3 as B2. Deposit syrupUSDC at 5.06% (share price 1.176549 USDC).

Additional risk vs B2: private credit default. Maple lends to leveraged crypto trading firms; the 135bp is a credit spread.

Correlation: a 57% ETH drawdown — the liquidation trigger — is precisely when those borrowers are most stressed and the redemption queue is longest. The ETH-price risk and the credit risk fire in the same tail event.

8. Path B4 — All-USDC → Maple syrupUSDC

The no-swap version of B3: borrow USDC directly, deploy to Maple.

8.1 Best USDC borrow venue — Aave v3 core, by elimination

Compound v3 USDC does accept wstETH (borrowCF 82%, liqCF 86%, supply cap 75,000 vs 39,254 posted), but sits at 89.01% utilisation against a 90% kink with slopeHigh = 360%.

Morpho’s adaptive IRM: AdaptiveCurveIRM targets 90% utilisation and ratchets the rate upward over days while above target. The 3.97% would not survive our $8.4M. Direct evidence: the Morpho wstETH/USDT market — identical collateral, identical 86% LLTV — sits at 92.1% utilisation paying 5.68%, 172bp higher purely from utilisation drift.

8.2 Sensitivity to Aave USDC utilisation

Maple deploy leg fixed at 5.06%:

The distance from “working” to “worse than doing nothing” is 0.08pp of utilisation. One point past the kink is a −2.30% absolute loss.

8.3 Observed utilisation history — 14 weekly on-chain samples, ~90 days

Utilisation note: balance-derived utilisation (totalDebt / totalAToken) runs up to 0.73pp above Aave’s internal virtual-balance measure. IRM parameters were verified identical across all sampled blocks for both stablecoin reserves (Uopt 92%, base 0, slope1 4%, slope2 100%), so the observed borrow APR is ground truth and utilisation is back-solved from it.

8.4 Why B4 is worse than the USDT paths

On raw spread cushion B4 is more robust than B1:

B4’s cushion is not thin — it is erasable in a single mechanical step. B1 requires Fluid and Aave USDT rates to genuinely converge, which is visible and exitable. B4 requires one basis point of utilisation drift across a threshold crossed twice in fourteen weeks, repricing instantaneously.

Verdict: reject. Removing the swap makes the position more fragile, not less — trading ~10bp of mean-reverting basis risk for a discontinuous ~110bp jump, on a trade whose entire upside is 35bp.

9. Rejected options

Not evaluated (outside a passive “deposit for a %” mandate): Pendle fixed-rate/PT markets · wstETH/ETH LP (LVR economics not assessed here) · restaking (EigenLayer, Symbiotic) · L2 venues. The staking-yield constraint is chain-independent, so loop economics do not improve on L2s, though borrow rates differ.

10. Master comparison

Basis: $24M equity = 10,503 wstETH · ETH $1,841.54 · wstETH $2,284.94 · Lido 2.21% · block 25,673,966.

Every computed cell below was independently re-derived from the stated inputs. Rates re-checked live at block 25,675,104 (+1,138 blocks): WETH borrow 2.1345% APY, USDT 3.7007%, USDC 4.0392%, Lido 7d SMA 2.2066%, fUSDT $131.13M, fUSDC $153.74M, syrupUSDC $1.0666B — all within rounding of the figures used.

Route B: net APY = 2.21% + 0.35 × (deployAPY − borrowAPY) · Route A: L × 2.21% − (L−1) × borrowAPY at L = 3.28x ($78.8M / $24M).

What would change the conclusion

1. Lido APR above ~3% — requires validator exits or a sustained fee-market recovery
2. WETH borrow rates fall — Aave core WETH at 80.7% utilisation against a 92% kink; a large supply inflow drops it toward the 0% base rate
3. Incentives return — a Merit or COMP campaign. Compound’s WETH comet already pays 0.19% on borrow
4. Spark or Compound raise caps — both are cheaper than Aave and cap-bound, not liquidity-bound. Spark’s WETH borrow cap moving 419,256 → ~500,000 opens ~$170M at 2.04%

Target: a ≥75bp staking-yield-to-borrow-cost spread. Currently +7bp on the venue with size.

11. Sources and contracts

Contracts (all reads at block 25,673,966)

Morpho market IDs (both wstETH collateral / USDC loan, 86% LLTV — see §8.1):

APIs (all accessed 2026-08-03)

Verification log

Discarded data: aggregator quotes fail to route large wstETH sells and return nonsense — Paraswap above 1,000 wstETH (−50% to −92%), CoW at ≥24,000 wstETH (a 24,000 quote returned 1,618 WETH, −9457bp). Validate any single large quote against the size curve before using it as a cost. DefiLlama’s chartLendBorrow endpoint is paywalled; utilisation history was sampled directly on-chain instead.

Unit convention: bp for rate spreads and execution costs · % for levels and price moves · pp for gaps between two percentages.