Weekly Intelligence — Week of September 28–October 4, 2026

1. Staking health

The question this week is not “how healthy is staking” but “how does a healthy network absorb a security event” — and the week answered it in public. MetaMask's precautionary exit of roughly 17,000 validators (523,000 ETH) pushed the exit queue from ~166,000 ETH (Sep 29) to a peak above 850,000 ETH (Oct 2), the largest backlog since December 2025, with the exit wait stretching to roughly 14–15 days. Underneath the spike, the entry queue held 2.48M ETH with a 27.7-day wait — the Sep 28 Ebunker board, carried — so demand to enter still ran well above the queue to exit even at the peak week (the 850,000 peak against 2.48M waiting to enter is roughly a 3× imbalance in the protocol's favor).

The board itself did not refresh, and that is the disclosure rather than a gap. Ebunker's October weekly was not obtainable inside the window, and no other dated network read inside September 28 – October 4 published an active-balance, validator or supply figure we can stand behind — so the whole staking register stands at the Sep 28 print: 43,681,037 ETH staked (43.68M), a 36.19% staking ratio, 906,000 active validators (last obtainable read Sep 20/21) and a 2.18% network base APR. Every one of those four numbers carries its own date this edition, and no week-over-week change is claimed for any of them. The carried board is the honest counterpart to an edition that otherwise printed a nine-figure security event: the network's admission and exit machinery processed the largest security-driven exit in a year without a hiccup, which is precisely what the churn limit is for, and the staking register simply sat still while it happened.

Source: queue path via Crypto Briefing and COIN360 (Oct 2); entry-queue volume and pre-event exit wait per the Ebunker Ethereum staking weekly (Sep 28, 2026); carried staking register (staked balance, ratio, validators, base APR) per the same Sep 28 board; validator delta vs the carried Sep 20/21 read.

2. stETH: supply and peg

The peg spent the week inside a ±16bp band on the single-read basis: from +5.4bp (Sep 27) through a week-high +15.2bp (Sep 28) to a close of −6.4bp (Oct 4; ratio 0.99936). A validator-exit week that once would have been framed as a peg stress test instead printed the quietest ratio path in a month — the exit was an operator security drill, not a redemption run, and the ratio behaved like it. One honesty note: the Sep 27 point re-read as a 5bp premium this week against edition six's printed 5bp discount; the flip sits inside the feed's re-print noise and both prints are disclosed (data report, section 3). Supply: still no fresh dated print — the 9,659,594 stETH figure is 26 days old, and the data report this week suspends the supply×price cross-check rather than run a stale input through a live comparison. In its place the same quantity is derived from two in-window prints: Lido's $26.51B TVL at the Oct 4 ETH print implies 9.84M ETH under management, or roughly 22.5% of the 43.68M ETH staked network-wide (both DERIVED on inputs dated inside the window).

Source: stETH/ETH series, DefiLlama coingecko feed, same-session prints; supply per CryptoCompare via TickerReport (Sep 9, 2026); TVL per DefiLlama protocol history (Oct 4 point); share-of-stake DERIVED on the TVL and Sep 28 staked-total prints.

3. DeFi / LSDfi

Third straight fully fresh board, and the flattest yet: six of eight rows moved less than one percent. Lido $26.51B (+0.1%), Aave $19.10B (−0.7%), Binance Staked ETH $10.02B (−0.2%), Morpho $11.31B (+2.5%, the only real mover — and the only row that grew through a DeFi-wide selloff session), EigenCloud $7.04B, Sky $6.07B, ether.fi Stake $5.18B, Rocket Pool $1.40B. Ethereum chain TVL slipped −0.5% to $53.28B. The Sep 29 DeFi selloff that took LDO down 10.5% in a session left no visible mark on any TVL row — governance-token prices repriced; deposit capital did not move. That is the cleanest possible separation between token beta and protocol usage, and it recurs every time this column checks for it.

Two readings ride on that flat board. First, the compression is getting tighter, not looser: this week's maximum single-row move (Morpho, +2.5%) is the smallest “only mover” spread this column has recorded, which is what a maturing sector looks like on a quiet tape. Second, none of Lido's liquid-staking competitors gained share during the one week a large chunk of Lido's own stake was withdrawing — the exits ran through the beacon-chain queue, not through the TVL ledger, so the event that dominated the news column left the sector board almost untouched. The board is, increasingly, the boring half of the story, and boring is the property depositors pay for.

Source: api.llama.fi protocol-history reads, one session, Oct 4 points; UNI/LDO series, DefiLlama coingecko feed.

4. Competition

The treasury bid kept its cadence: Bitmine's September 28 SEC exhibit listed 6,001,302 ETH — across the six-million mark, roughly 11.6% of the network's staked ETH (DERIVED: its 5,067,309 staked against the 43,681,037 ETH Sep 28 board), with 5,067,309 ETH staked and its own 7-day yield print at 2.62%, flat on the week. No post-window treasury update is used in this edition — Bitmine's next weekly print lands after October 4, so it stays in the next edition rather than being run early. The operator side of the competitive board printed nothing new this week: no fresh provider APRs were reachable, so the corridor's carried rows stand — Lido user-net 2.21–2.24%, frontend 2.40%, rETH 2.19%, all labeled with their print dates. The one competitive datum with a date: Lido's exit-queue handling — the ad hoc reserve fund of more than 6,750 stETH, the 45-day re-entry estimate, the no-action notice to stETH holders — is a live demonstration of what diversification across ~900 operators buys when one of them exits ~523,000 ETH at once.

Source: Bitmine SEC exhibit (Sep 28, 2026) via MEXC research; network staked total per the Ebunker weekly (Sep 28, 2026); Lido disclosures (Sep 30, 2026) via Blockhead; carried rate rows per their dated sources in the data report.

5. Outlook — and edition six's four calls, graded

Edition six published four watchlist items. Three can be settled this week; the fourth resets.

Watchlist for the next edition, four items: (1) the exit-queue drain — the final MetaMask validators were scheduled to enter the exit process by October 7, so the queue should fall steadily from its 850,000 peak; a stall past mid-October flips the story from security drill to something else; (2) the re-entry wave — 523,000 ETH returning through a 27.7-day entry queue; whether it re-stakes through Lido or waits is observable in the entry-queue composition; (3) the Ebunker series — whether the October print lands and grades the 2.18% step as step or regime; (4) the October 28 FOMC — the weak Oct 2 jobs report moved a hike from likely to unlikely; the meeting itself is the print that would move the carry tables.

Source: scorecard inputs as cited in sections 1–4 and the data report.

6. The security event, structurally

Worth separating what the MetaMask incident did and did not touch, because the two lists are the whole story. Touched: one operator's signing infrastructure; block-fee routing on 18 of 19 examined validators; roughly 0.36 ETH of rewards; the reward timing of every client whose validators exited (up to a 45-day round trip). Untouched: user withdrawal keys (MetaMask holds none); stETH redemption; the peg (inside ±16bp all week); the protocol's solvency; any other operator's validators. The attacker's take — under $1,000 — against a nine-figure defensive response is the cleanest illustration yet of where risk actually lives in liquid staking: not in the principal, which custody design protects, but in the yield's continuity, which only operator quality protects. Lido's buffers — operator spread, the 6,750 stETH reserve — exist for exactly this week.

The structural point is durable beyond this incident. Liquid staking's risk boundary is not “can the protocol lose my ETH” — it almost cannot, by construction — but “for how long can an operator interrupt my yield.” That second question is the one every prior operator-exit event (Kiln in 2025, the Consensys-name 2023 mistake) answered in days, and the one this week answered in a 45-day estimate. The peg holding inside a 16bp band while 523,000 ETH left is the market pricing exactly that: principal safe, yield timing at risk, and timing risk small enough to be absorbed by a reserve fund measured in thousands of stETH.

Source: incident detail per sections 1–2 of the news review; peg path per the data report; Lido buffer disclosures via Blockhead (Sep 30, 2026).

7. The macro leg moved first

The October 2 jobs report (+29,000 payrolls, 4.2% unemployment, −60,000 in revisions) did what three weeks of price action could not: it moved the rate path. October-hike odds collapsed from roughly two-thirds-to-three-quarters in late September to a hold probability near 72–77% by the print's close, with the October 28 meeting now the focal date. For the carry tables this site runs, the direction matters more than the level: the gap between staking's ~2.2% user-net yield and cash rates is widest precisely when hikes are the base case, and every probability point that migrates from “hike” to “hold” narrows the carry disadvantage without a single basis point of yield changing. The ETF complex −$138.1M week says institutional flows have not traded that repricing yet; the rate leg and the flow leg are running on different clocks, and the jobs report bought the market time rather than a bid.

Source: BLS Employment Situation (Oct 2, 2026); FedWatch pricing via China Securities Journal (Oct 2, 2026); Farside weekly net.

8. Where Lido sits after the week

Score the week structurally: TVL flat at $26.51B (+0.1%) through the sector's worst session in weeks; share of stake roughly 22.5% (DERIVED on in-window inputs, disclosed); the validator count flat on a carried board with zero stake disruption; one of its ~900 operators exited 523,000 ETH in a security drill and the peg, the TVL and the redemption machinery all printed “routine.” The things that would actually dent Lido's position — a redemption failure, a peg break, an operator loss of user principal — did not happen, and this week offered the sharpest test of that statement since the site began publishing. The carry-over risk is timing, not solvency: rewards resume on the chain's re-entry schedule, and the 27.7-day entry queue is the meter.

Source: TVL and ratio per the data report; incident figures per section 1–2 of the news review.

9. The trend this edition extends

Two of this edition's prints extend trends this column has tracked for weeks, and one breaks one. The TVL board's third straight fresh read extends the board's most useful property: comparability. The break: the ETF flow series, five-for-five positive last edition, five-for-five negative-or-flat this one — the series has now printed a +$830M two-week swing, and any analysis that treated the inflow week as a regime (ours included, cautiously framed) owes the outflow week the same skepticism. Flat price through both is the tell that neither leg is the marginal buyer the August tape made it look like. The carried staking board is the third thread: it breaks the run of fresh weekly prints this column has enjoyed since edition five, and it is the correct break — a register that says “no new data” is more informative than a register that invents one.

Source: series as cited in sections 1–5 and the data report.

FAQ

Related reading

• Lido & stETH Weekly Data Report — Week of September 28–October 4, 2026
• stETH Yield Update — Week of September 28–October 4, 2026
• The Exit Queue Shock: What 850,000 ETH Exiting Actually Means