Weekly Intelligence — Week of September 14–20, 2026
1. Staking health
Answer: participation rose, the validator count fell, and the yield the network pays is still below the rate the Fed pays banks.
Staked ETH reached 43,179,577 on the September 14 Ebunker weekly — about 35.77% of a 120.72M circulating supply, a rise of 0.22pp against our September 7 print of 35.55% on the same denominator. The same-day validatorqueue read puts it at 43.2M and 35.39%, on a denominator about 1.8M ETH larger; we print both because they are not comparable and we will not blend them.
The number that deserves more attention than it got is the validator count: 906,000 active on the latest available read (validatorqueue), against 911,414 on Sep 13. It fell 0.6% in a week when staked ETH rose. That is the first visible effect of Lido Core's CMv2 consolidation — the same 2,048-ETH balances carried by fewer keys. The yield consequence is real but small: the network staking APR reads 2.57% (Ebunker, Sep 14) against 2.59% on validatorqueue's Sep 10 read. Against a policy rate the Fed now projects at 4.1% through 2027, the shortfall is roughly 153bp (DERIVED).
Net issuance was +20,397.71 ETH over the week, which annualises to about 0.88% of supply (DERIVED: 20,397.71 ÷ 7 × 365 ÷ 120,716,586) — a touch above the 0.86% we had carried since September 2, and the first fresh read on that metric in three weeks.
Source: Ebunker Ethereum staking weekly (Sep 14, 2026); validatorqueue.com via Tremplin (Sep 20) and CoinOKC (Sep 13); churn limit per Ethereum protocol parameters.
2. stETH: supply and peg
Answer: the peg passed its second live stress test and closed the week above par; the supply figure is the weakest input we publish.
On the daily same-source pair, the stETH/ETH ratio ran 1.00052 (Sep 13) → 0.99906 (Sep 16, a 9bp discount on the Fed day) → 1.00049 (Sep 20, a 5bp premium) Three things are worth saying about that path. The discount was the widest of the week and arrived on the single most macro-sensitive session of the quarter — and it was 9bp, not a dislocation. It closed within three sessions. And the recovery overshot to a premium, which is what a wrapper with a working redemption path looks like when leveraged holders have finished unwinding.
Supply is the weak leg. Our last dated figure remains 9,659,594 stETH from September 9, eleven days old, and it no longer reconciles with TVL: at today's price the $25.72B TVL implies about 9.81M, a 1.6% gap. Aggregators disagree with each other as well — BingX prints 9,657,537 for Sep 7, on our basis; LBank prints 9.43M for Sep 18, which would widen the gap to 4.5%. We could close the gap by adopting either figure. We are not going to, because neither is a dated read we can defend.
Source: stETH/ETH ratio from the DefiLlama coingecko feed, same-source same-day prints, read Sep 20; supply print per CryptoCompare via TickerReport (Sep 9); TVL from api.llama.fi/tvl/lido (Sep 20); aggregator prints accessed Sep 20.
3. DeFi / LSDfi
Answer: the whole board reflated with the price, and the fresh data lets us drop the asterisk.
For the first time since the founding edition, every row of our sector board is a direct read rather than a carried print. Ethereum's chain TVL stands at $52.58B (DefiLlama, Sep 20), up 7.4% from the $48.95B September 7 board; the next-largest chain, Solana, is $6.52B. Inside it, Lido is $25.72B, Aave $18.99B and Morpho $10.73B — all three at all-time highs for our series, all three DERIVED changes of +5.6%, +5.1% and +11.1% against the September 7 board.
The honest caveat is that this is mostly a price move, not a deposit move. TVL is denominated in dollars; ETH ended the window up 3.8% (after trading as much as 4.9% below the Sep 13 close at mid-week). The one flow reading that is not price-driven is the entry queue: 2,480,000 ETH still waiting to activate, a queue that has barely moved in two weeks even as the price swung about 9% peak-to-trough. Demand to lock ETH up has been insensitive to the tape, in both directions.
Source: DefiLlama API direct reads (api.llama.fi/tvl/lido, /tvl/aave, /tvl/morpho, /v2/chains), Sep 20, 2026; queue per validatorqueue via Tremplin, Sep 20.
4. Competition
Answer: Lido's lead held through a week in which its nearest custodial rival grew faster, and the two structural Lido events of the week both cut operating cost or governance surface rather than chasing deposits.
| Board | TVL, Sep 20 | Sep 7 board | Change |
|---|---|---|---|
| Lido | $25.72B | $24.35B (Sep 15) | +5.6% |
| Binance Staked ETH | $9.81B | $9.15B | +7.2% |
| ether.fi Stake | $4.91B | not carried | — |
| Rocket Pool | $1.37B | not carried | — |
Lido is 2.62× Binance Staked ETH (DERIVED), a ratio that has widened slightly in Lido's favour on the prior board. The four liquid-staking protocols on this board hold $41.81B between them, and Lido plus Binance Staked ETH account for 85% of that (DERIVED). But the more interesting competitive fact is structural. On Sep 17 Lido began migrating roughly 8.4M ETH onto CMv2, whose 34 curated operators all carry ETH bonding requirements; on Sep 18 it retired Simple DVT, a module that existed specifically to widen its operator base. Read together, Lido spent the week making its validator set narrower and better capitalised — the opposite direction from the decentralisation narrative its competitors sell against. Whether that trade is worth it is the subject of this week's deep dive.
Source: DefiLlama API direct reads, Sep 20, 2026; Lido Core announcement (Sep 17) via Crypto Briefing and CoinPost; Simple DVT retirement (Sep 18) via Edgen.
5. Outlook — and edition four's four calls, graded
Edition four published four watchlist items, each with a number, a direction and a deadline. All four can be settled this week:
| Call (edition 4) | What the data did | Verdict |
|---|---|---|
| Exit-queue drain rate: “2.656M ETH should fall toward zero inside ~2 weeks; a stall flips the story” | Sep 13 read: exactly zero ETH. Sep 20 read: 182,352 ETH. | Held on its own terms — and it is the item that forced the check catching our own misdated source (see the data report's ledger) |
| Peg discount: “0.9987 is 13bp; watch whether it closes or widens” | Widened to 9bp on the Sep 16 chart-basis print, then closed to +4.9bp by Sep 20 | Held — transient, as a functioning redemption path implies |
| FOMC Sep 15–16: “the macro leg of every carry table reprices this week” | A 25bp hike on Sep 16; the cash-versus-staking gap widened, as the pre-committed translation table said it would | Arrived — direction right, size not ours to call |
| LDO supply basis: “four weeks conflicted; resolution would let us print a cap without an asterisk” | 834,151,322 tokens (Sep 17 data set) against our carried 895.8M | Resolved — downward, by 6.9% |
Watchlist for the September 21–27 edition, four items again: (1) the supply print — a fresh dated stETH supply figure would either close the 1.6% cross-check gap or confirm that supply has grown into it; (2) CMv2 migration pace — Lido has published one consolidation. A second and third in the next fortnight would confirm the Q1 2027 schedule; a month of silence would not; (3) the LDO supply basis, second source — 834.2M is now a single-source figure, and our own rule says a single source is a lead, not a fact; (4) the Fed's October meeting — the SEP's 4.1% median is a projection, and the next print of the policy rate is the thing that would move the 153bp staking gap again.
Source: scorecard inputs as cited in sections 1–4 and the data report; FOMC decision and SEP, Sep 16, 2026; CMv2 timeline per Lido's Sep 17 announcement.
6. Net issuance: the quiet part of the supply story
The headline supply print is stale, but the flow behind it is fresh, and the two do not yet reconcile. Net issuance ran +20,397.71 ETH over the week, which annualises to about 0.88% of supply (DERIVED: 20,397.71 ÷ 7 × 365 ÷ 120,716,586) — a touch above the 0.86% we had carried since September 2, and the first fresh read on that metric in three weeks. At that rate the network mints roughly 2,900 stETH a day, or about 20,400 stETH across the window.
That is the number that should be closing the supply–TVL gap, and it is not. The gap we carry open is 150,195 stETH (DERIVED: implied 9,809,789 from $25.72B TVL minus the dated 9,659,594). Eleven days of 2,900-ETH issuance adds only ~32,000 stETH, an order of magnitude short of what the TVL-implied supply says is missing. So the gap is not issuance; it is the dated print being too old to track a market that moved 9% in a week. We say so plainly because the alternative — that 150k stETH evaporated — is not compatible with any on-chain reading.
Source: net issuance and the 120.72M denominator per Ebunker Ethereum staking weekly (Sep 14, 2026); dated supply per CryptoCompare via TickerReport (Sep 9); TVL-implied supply from api.llama.fi/tvl/lido (Sep 20).
7. CMv2, mechanically: why keys fall while stake rises
Section 4 noted the structural fact; the mechanism is worth a section of its own. Ethereum's Pectra fork let 0x02 validators hold up to 2,048 ETH instead of 32, so the same stake can be secured by one-sixtieth the number of keys. Lido's CMv2 migration, which began September 17 and moves roughly 8.4M ETH onto 34 curated operators (all carrying ETH bonding requirements), simply consolidates those keys. The stake does not move; the key count does. That is why the validator set fell 0.6% in a week when staked ETH rose.
The trade-off is the point. Consolidation lowers Lido's operating surface and raises its capital efficiency, but it narrows the operator set at exactly the moment rivals sell decentralisation as their differentiator. Simple DVT, the module Lido retired September 18, existed specifically to widen that set; retiring it reverses the direction. Whether a narrower, better-capitalised validator set is worth the narrative cost is the subject of this week's deep dive, and we are not prejudging it — but the mechanism that produced the −0.6% key count is now documented, not inferred.
Source: Lido Core CMv2 consolidation announcement (Sep 17, 2026) via Crypto Briefing and CoinPost; Simple DVT retirement (Sep 18) via Edgen; Pectra 0x02 mechanics per Ethereum protocol documentation.
8. The macro overlay: a hiking Fed and a widening staking gap
Every carry table on this site has a macro leg, and this week it repriced. The FOMC hiked 25bp on September 16 to a 4.00% policy rate, with a Summary of Economic Projections median of 4.1% through 2027. The network staking APR, by contrast, reads 2.57% (Ebunker, Sep 14). The shortfall against the policy rate is therefore 143bp (DERIVED: 4.00 − 2.57); against the SEP median it is 153bp (DERIVED: 4.1 − 2.57).
This is the worst the staking-vs-cash gap has looked since this column began, and the reason is two-sided: the reference rate rose while the network APR fell 2bp. For a stETH holder the consequence is not a forced sale — the redemption path is open and the peg is above par — but the opportunity cost of holding staked ETH just widened by exactly the 25bp the committee delivered. That is the whole macro story in one figure, and it is why section 4's structural Lido events, not the rate, are the week's real news for stakers.
Source: FOMC decision and SEP (Sep 16, 2026); network staking APR per Ebunker Ethereum staking weekly (Sep 14); all gaps DERIVED from those two.
9. Where Lido sits in the liquid-staking field
A share of stake is the cleanest single summary of competitive position, and it held. Lido controls 22.4% of all staked ETH (DERIVED: 9,659,594 ÷ 43.179M), and 2.6× the next-largest liquid staking token, Binance Staked ETH (DERIVED: $25.72B ÷ $9.81B). The four liquid-staking protocols on our board — Lido, Binance Staked ETH, EtherFi and Rocket Pool — hold $41.81B between them, and Lido plus Binance account for 85% of that (DERIVED).
The number to watch from here is not the share but its denominator. Staked ETH rose to 43.18M this week while Lido's own supply print sat frozen at 9,659,594 from September 9. If the next dated stETH print confirms supply growth, Lido's share can fall even as its TVL rises — a quietly important distinction that the 22.4% figure papers over. We flag it because the share is the most-quoted Lido metric and the one most likely to be read as a trend it is not yet one.
Source: stETH supply per CryptoCompare via TickerReport (Sep 9); total staked per Ebunker weekly (Sep 14); peer TVL per DefiLlama API direct reads (Sep 20).
Sources
- Staking data — Ebunker Ethereum staking weekly (Sep 14, 2026)
- Queue reads — validatorqueue.com via Tremplin (Sep 20, 2026)
- Validator count, Sep 13 — validatorqueue.com via CoinOKC (Sep 13, 2026)
- Lido TVL — DefiLlama API direct read (Sep 20, 2026)
- Chain and peer TVL — DefiLlama direct reads: chains, aave, morpho, binance-staked-eth, ether.fi-stake, rocket-pool (Sep 20, 2026)
- CMv2 consolidation — Crypto Briefing (Sep 17, 2026)
- Simple DVT retirement — Edgen, citing the Lido governance forum (Sep 18, 2026)
- FOMC decision and SEP — Federal Reserve (Sep 16, 2026)
- LDO circulating supply — Analytics Insight (Sep 18, 2026)
- ETH / stETH daily series — DefiLlama coingecko feed (read Sep 20, 2026)
FAQ
Is a 35.77% staking ratio comparable with the 35.55% you printed on Sep 7?
Yes, and that is the change this week. Ebunker's 35.77% implies a 120.72M circulating denominator, the same one behind the Sep 7 print. The 35.29% we published on Sep 15 was on a ~122.1M denominator and is not comparable with either; we label that rather than blend it.
Why is the validator count falling while staked ETH rises?
Consolidation. Ethereum's Pectra fork allows 0x02 validators to carry up to 2,048 ETH instead of 32, so the same stake can be secured by fewer keys. Lido's CMv2 migration began this week and projects roughly a third fewer validators; the network count fell 0.6% in the same seven days.
Does a 9bp peg discount mean stETH is broken?
No, and the sequence matters more than the number. The discount appeared on the Fed decision day, lasted a session, and had closed to a premium within three. What would be concerning is a discount that persists while redemptions are available — that has not happened in six weeks of tracking.
Is this financial advice?
No. Informational analysis only.