Weekly Intelligence — Week of August 24–30, 2026

1. Staking. How healthy is Ethereum staking?

No fresher verified read than the Aug 13 anchors carried forward: 41.89M ETH staked, a 34.7% ratio, ~789K active validators, 0.86% inflation. One third-party table this week cited ~39M ETH staked and a 900+ operator set — different vintages and counting bases, flagged in our data report rather than averaged. The structural picture is unchanged: record saturation, dilution grinding on. What did change is the macro backdrop the staking trade lives inside — see question 4. The one-line summary of the week: staking's fundamentals did not move; the ground under the trade did.

One carried number deserves more airtime: validator headroom. 41.89M ETH staked, in a design that organizes stake into 32-ETH slots, corresponds to roughly 1.31M validators' worth of stake (DERIVED: 41.89M / 32) against ~789K active — about 60% of that slot-equivalent earning in the active set (DERIVED). The design consequence: the network absorbs a large exit wave by thinning everyone's rewards rather than by breaking, which is precisely the property this month is testing, and the reason an exit queue can hold millions of ETH (as September 2025 showed) without the chain stopping.

2. stETH. Supply, rate and anomalies

Supply is this week's honest problem: lido.fi's counter reads 9.63M ETH (Aug 27) while an aggregator table shows 8.86M ETH (undated). We publish the conflict and withhold the WoW delta — a 9% spread between sources is not a rounding difference, and guessing which is right would defeat the point of a verification-first column. On the rate: the last verified APY print remains 2.17% (Aug 15); the frontend's rolling APR now reads 2.4% (Aug 31) — different metric, not a fresh signal. Anomalies: none dated inside the window; the peg held through the week's worst risk-off day, which is the anomaly test that matters. Supply × price now reads $24.1B against a TVL of $18.7B — the two stopped reconciling once we corrected the ETH price, and we are leaving the gap visible rather than picking the number that looks better. Same-day reads are scheduled for next issue.

3. DeFi/LSDfi. What changed inside the window?

The freshest aggregator table sharpened the map: Lido at $18.7B TVL, the seven non-Lido LST protocols together at 5.27M ETH ($11.11B), with Binance's wBETH alone at $7.80B — about 70% of the non-Lido field, and Lido plus wBETH together ~89% of the $29.82B board. The 30-day flows inside that table tell a directional story: Rocket Pool -22.75%, mETH -20.58%, StakeWise -8.46%, while cbETH +4.13% and wBETH roughly flat. The decentralized tier is bleeding toward zero, and the market is consolidating into one protocol plus one exchange. For LSDfi strategies, the fee-honest net table (Lido ~2.16% vs cbETH ~2.12% net) is now live in our data report — the first comparison where fees are counted for everyone.

The consolidation arithmetic, stated once: if Lido plus wBETH are ~89% of the $29.82B board, the remaining five tracked protocols share roughly 11% of it (DERIVED) — on the order of $3.3B split five ways (DERIVED on the printed table). The 30-day flow column says where even that remainder is drifting: cbETH's +4.13% is the only positive row, and it is custodial-adjacent. For the LSDfi build-out this column tracks, a two-protocol board changes the design brief: strategies that assume many competing yield sources must now assume one dominant source with fee-setter power. That is not a moral judgment; it is a planning input.

4. Competition. Where does Lido stand after the week's macro turn?

Position unchanged, environment worse. Lido remains #1 at $18.7B TVL; the custodial bloc (wBETH $7.80B) remains the true #2. The week's macro turn — September hike odds to 56–60%, the 2-year Treasury at 4.34% — compresses every argument for holding a ~2.2%-net-yield asset, and hits no-cash-flow governance tokens hardest: LDO fell 12% on the week, and did so while Bitcoin was still closing August up ~26% month-to-date. Rate parity with rETH (last same-source read) held until the series went stale; the new cross-provider table shows Lido's net ~2.16% above rETH's ~1.72% net — but on a re-baselined metric we will hold constant before claiming a trend. The strategically important line: in a hiking regime, Lido's moat is integration depth and liquidity, and neither moved this week.

On competition specifically, the macro turn reprices the whole field, not one protocol. A rising cash yield compresses the relative premium of every liquid staking token simultaneously; what it does not change is relative share, which moves on fees, integrations and trust — slower variables. The week's evidence cut that way: no share-shifting event inside the window, plenty of relative-yield repricing. Treating those two clocks separately — a fast clock for rate-sensitive valuation, a slow clock for structural share — prevents the most common weekly-intelligence error, which is reading a rates move as a competitive defeat.

The LDO print is the fast clock made visible. The week's -12% took the token from $0.4007 — above the $0.29–0.37 band it has held all month — to $0.3527, back inside the band roughly 5% below its ceiling (DERIVED). Nothing about Lido's fee engine moved: the same 10% fee on the same gross yield, the same operator set. What moved is the discount rate applied to a no-cash-flow claim in a 4.34% two-year world. The slow clock — share, fees, integrations — printed nothing this week, and the scorecard below grades accordingly.

5. Outlook. What to watch next week?

• The September 15–16 FOMC is now the singularity: a 56–60% hike probability, a chair who refuses to guide, and every LST valuation keyed to the outcome.
• Same-day supply re-verification — our own checklist item, to repair the failed cross-check.
• ETF flow direction — desk commentary flagged outflows limiting the weekend rebound; a reversal would be the first genuinely staking-positive institutional signal in weeks.

Grading your own calls a week later is the least glamorous and most valuable section in this column. The discipline is mechanical: each outlook item from the founding edition is restated exactly as published, marked against what the window actually printed, and scored held, direction-right or contradicted — no partial credit for being "basically right," no silence about calls that failed. The point is not scorekeeping for its own sake. Outlook sections across this industry are overwhelmingly unaudited, which makes them free to be wrong; once a column commits to grading, every future call is written with the knowledge that it will be read back. That asymmetry is the product.

Scenario framing for the singularity, without pretending to know the outcome: a 25bp hike lifts the cash leg of every carry table by 25bp (DERIVED), taking the roughly 2.2-point gap toward 2.4; a hold freezes it; a surprise cut — not the base case at 56–60% hike odds — would compress it and re-rate every no-cash-flow token in the sector at once. The useable insight is the asymmetry of attention: the hike path is largely priced at those odds, so next week's grading question is not "which way did rates go" but "did the outcome deviate from the odds that were already printed."

6. Scorecard: grading edition one's calls

The founding outlook made four trackable calls. Grading them against the window:

Scorecard: 1 held, 1 direction-right, 1 unchanged/untested, 1 re-baselined. Nothing contradicted. The grading is ours and deliberately harsh where data went stale — the scorecard's job is to make our own misses expensive to hide.

7. How to use this column

Three layers, in reading order: the five answers give you state; the watchlist in the founding edition gave you triggers; this scorecard gives you accountability. Next week the cycle completes — the FOMC outcome gets graded, the supply conflict gets resolved or escalated, and the parity call gets its same-source test. The five questions never change, so every edition stays comparable; the scorecard makes sure we pay for our forecasts in public.

FAQ

Related reading

• Lido & stETH Weekly Data Report — Week of August 24–30, 2026
• Lido & stETH Weekly News Review — Week of August 17–23, 2026 (founding edition)