The Ethereum Foundation’s 6 November 2026 post is a developer call, not a mainnet clock. Marius van der Wijden writes that Glamsterdam, the next Ethereum protocol update, is still on testnets, that EIP-8037 and EIP-8038 will break eight gas invariants some contracts still assume, and that All Core Devs (ACD) will not fork Hoodi until those issues are resolved. Hegota work continues in parallel.
If you trade ether candles, this is not a range break. If you ship a smart contract, a relayer, or a wallet that guesses gas, it is homework. The Foundation even published a repository of the eight invariants and asked auditors to run their clients’ code against it.
Ether is still the tape around the software. Our CoinGecko snapshot at has ETH at $2,240.29, up 0.21% over 24 hours, with Bitcoin at $75,906 (down 0.07%) and Solana at $90.21 (up 1.11%). Global crypto market cap sits near $2.57 trillion, with Bitcoin dominance near 59.23%. Coinbase BTC-USD last traded at $75,806.50 at , inside a 24-hour venue range of $74,722 to $76,288.14. Quiet tape is not a fork date.
What happened
The Foundation post, filed under Research & Development and dated 2026-11-06, says Glamsterdam is being tested on testnets. During those tests, Chainsecurity found a few projects that would be vulnerable after Glamsterdam because they rely on invariants that EIP-8037 (state-creation gas cost increase) and EIP-8038 (state-access gas cost update) will break. Those EIPs rework how much gas is charged for certain operations and how gas is charged for state. The public EIP-8037 page we fetched still shows Last Call with a 2026-11-01 deadline and an abstract that raises state-creation costs, introduces CPSB (cost per state byte), and meters state creation separately so large deployments are not capped by a single transaction gas limit.
Van der Wijden points readers to Chainsecurity’s 13 October 2026 note, Glamsterdam Hard Fork: When Spent Gas Comes Back, and to a repricing tool. He then states the process decision in plain language: based on those findings, ACD has decided to delay the hardfork process and not fork Hoodi until the issues are resolved. Hegota is not paused. The post asks contract developers and auditors to check whether they rely on any of the eight invariants, and it points to the GitHub repo ethereum/glamsterdam-gas-invariants. If a project is affected, the contact is glamsterdam at ethereum.org, not a public GitHub issue.
The repo README (default branch master, created 2026-11-01, last push 2026-11-06 09:21 UTC) is the operational document. It lists eight invariants, INV-1 through INV-8, split into accounting changes from EIP-8037’s new gas tracking and price changes from the new opcode costs. It says most findings so far reference INV-4 and INV-7, so those should be checked first. It also publishes a measured cost table from EELS on a pinned Amsterdam spec, with the caveat that the numbers change if the spec changes. Examples: a cold SSTORE to a new slot moves from 22,100 gas on Osaka to 110,020 under Amsterdam (2,100 + 10,000 + 97,920 state gas); a value transfer that creates a new account moves from 25,000 to 183,600 state gas plus execution; code deposit moves from 200 gas per byte to 1,530 per byte as state gas. A separate state-gas reservoir exists only when tx.gas is above 2^24 (16,777,216). Below that, all state gas comes out of gasleft().
Context
This is not the first Hoodi delay note, and it should not be filed as a reprint of 22 October. That earlier Foundation post postponed Hoodi’s Glamsterdam activation (the 26 October epoch 132352 clock). Client tags then turned the fork off on Hoodi. Today’s post is a different artifact: it names the invariant class, ships a checklist repo, and asks the next ACD calls for a pragmatic decision on whether EIP-8037/8038 need a small spec change. Recycling the October postpone as this story would miss the new evidence.
Chainsecurity’s October 13 write-up is the technical companion, not a second rumor mill. It argues that gas refills and repricing can let failure handlers complete where they once reverted, so existing gas checks can silently change meaning even if contract bytecode is unchanged. That matches INV-1, INV-3, and INV-6 in the Foundation repo: out-of-gas is no longer a reliable “the whole transaction died” signal; a reverting call can hand storage gas back; leftover 1/64 gas after a failed call can be enough to write a permanent failure flag. INV-8 is the wallet-shaped one: a 7702 delegation that runs out of gas can roll back while the transaction still mines, so a mined tx is not proof the delegation stuck.
Fee context still matters because Glamsterdam is a fee-market rewrite, not a branding exercise. DefiLlama’s Ethereum fees overview, fetched at from api.llama.fi/overview/fees/ethereum, prints about $18.71 million in trailing 24-hour fees (total24h 18,707,978.84). The chart’s last complete UTC day in that JSON (2026-11-05 00:00 UTC) is about $18.65 million. Those are usage prints, not a forecast of post-Glamsterdam costs. Ethereum still collects real gas fees while the next fork spec is stress-tested. Show me the fee, not the roadmap.
Ethereum’s history is a chain of upgrades that changed gas meaning. Berlin (EIP-2929) already repriced state access in 2021, and EIP-8038 says the state has grown enough to do it again. Traders who only watch the Ethereum news headline for a date will overfit. The useful object this week is the invariant list, and whether ACD treats INV-4/INV-7 reports as a spec tweak or as “fix your contracts.” For how we timestamp a tape like this, use the desk’s market snapshot guide. An audit that never re-ran gas assumptions after a repricing EIP is the failure mode the Foundation wants to catch before Hoodi, not after mainnet.
Our read
Named stance: this post is an invariant hunt that keeps Hoodi paused. It is not a mainnet activation, not a quiet “gas goes up a bit” note, and not a reason to mark a Glamsterdam date on a trading calendar. Hegota continuing in parallel is the Foundation’s own split: one fork’s gas accounting is blocked on contract-level assumptions, the other workstream is not waiting.
I care about this because fee markets are where Ethereum actually lives. If INV-1 and INV-6 are common in production, a failed inner call can commit a “done” flag that used to revert, which is a user-funds and settlement bug, not a gas-golf nit. If INV-4 and INV-7 dominate, as the README says they do in the Sourcify sweep, the pain is hardcoded gas in contracts, signed user operations, and transfer/send stipends that cannot create a storage slot at the new prices. That is boring until a relayer bricks a first-time recipient.
Falsifiable claim: by 2026-11-21 23:59 UTC, the 6 November Foundation post and the ethereum/glamsterdam-gas-invariants README will still present Hoodi Glamsterdam as delayed until the invariant issues are resolved, or a later Foundation/ACD note will say the same in writing. If ACD or the Protocol Cluster publishes a replacement Hoodi Glamsterdam epoch or UTC timestamp that is scheduled to activate before that deadline, and does so without stating that the eight invariants are resolved or no longer blocking, this read is wrong.
What to watch next
First, the next All Core Devs notes. The README says the repo exists so ACD can have a well-founded discussion of impact, “depending on the response we might decide to slightly change EIP-8037/8038 or not.” A spec tweak, a “contracts must migrate” line, or a new Hoodi epoch are three different outcomes. Only the written notes count.
Second, whether INV-4 and INV-7 stay the modal findings, or INV-1/INV-8 start showing up in mail to glamsterdam at ethereum.org. The Foundation asked for private email with a working exploit, not a public issue, so the public signal will be delayed by design. Treat Twitter screenshots as noise.
Third, client binaries. A Hoodi date that is not in an EF post or an ACD note is not a date. Watch execution-client tags the way we watched Besu 26.11.0: patch versus fork clock. Hegota progress is a separate checklist. It does not secretly unpause Hoodi.
Fourth, the tape. ETH at $2,240 on this snapshot is a grind, not a decisive range break. If ether rips on a rumor of a Glamsterdam date, check whether that date is in the Foundation post. Today it is not.