Ethereum Tests Glamsterdam Today. Its Biggest Upgrade Since the Merge Has No Mainnet Date
Ethereum’s Glamsterdam upgrade forks the Sepolia test network this morning at 9:53:36 a.m. ET, and the test network is the entirety of what has been scheduled. There is no mainnet date. The Ethereum Foundation’s own announcement says it in as many words: “Hoodi and mainnet activation dates have not yet been decided”, and “This announcement covers Sepolia; it does not schedule a mainnet upgrade.”
A test network fork is worth attention anyway, because Glamsterdam carries more core EIPs than any network upgrade since the Merge, and one of its two headline changes has been a live proposal since June 2024.
The fork, the time, and the two layers it combines
Glamsterdam is a single name for two coordinated upgrades: Amsterdam on the execution layer and Gloas on the consensus layer. On Sepolia it activates at epoch 353,024, slot 11,296,768, Unix timestamp 1791294816, which is 13:53:36 UTC and 9:53:36 a.m. ET today, Tuesday October 6 2026. The consensus layer’s activation epoch and the amsterdamTime value in the execution layer’s own Sepolia config land on the same second.
The scheduled contents run to 18 core EIPs, plus five networking EIPs and two informational ones. That is the largest core list since the Merge. Dencun, Pectra and Fusaka are counted from the core list in each one’s hardfork meta EIP, at nine (mainnet March 13 2024), ten (May 7 2025) and nine (December 3 2025). Shapella in April 2023, the fourth upgrade since the Merge, has no meta EIP, and its five are execution-layer EIPs only, a different basis from the other three.
Glamsterdam’s 18 scheduled core EIPs against nine to ten in each of the three upgrades before it.
Those four lists shipped. Glamsterdam’s is scheduled rather than shipped, and EIP-7773, the meta EIP that governs its contents, is still at Review status, so the list can change.
What ePBS and block-level access lists actually change
The first of the two headline changes is EIP-7732, enshrined proposer-builder separation, usually shortened to ePBS. Most proposers today do not build their own blocks. They hand construction to a third-party builder, and the swap runs through trusted middleware: the proposer signs a blinded block and, in the EIP’s term, a trusted party substitutes the real payload. The motivation section calls what replaces that a “trust-free fair exchange”. The proposer commits to a builder’s bid, the protocol deducts the promised payment from the builder’s beacon chain balance, and a payload timeliness committee of validators attests to whether the builder revealed the payload on time.
The same EIP splits consensus validation from execution validation, so a validator runs only the consensus state transition inside the four-second attestation window and can check execution and data availability across the rest of the slot.
The second is EIP-7928, block-level access lists. A block now carries an enforced list of every account and storage slot it touches along with the resulting values. In the EIP’s own words, that enables “parallel disk reads, parallel transaction validation, parallel state root computation and executionless state updates”. Execution cannot be parallelised unless you know in advance what each transaction will touch, and the same list allows “state reconstruction without executing transactions”.
Underneath them sits a round of gas repricing. EIP-8037 separately meters the cost of creating state, EIP-8038 updates state-access costs, and further EIPs cover intrinsic transaction gas, calldata and block gas accounting. The Foundation’s warning to developers is blunt: “Contracts that rely on fixed gas stipends, hardcoded gas limits, or assumptions about remaining gas may need changes.”
The six-second blocks are not in this upgrade
EIP-7782, “Reduce Block Latency”, would cut the slot time from 12 seconds to 6, halving confirmation times and shrinking an epoch from 384 seconds to 192. Its status in the ethereum/EIPs repository is Draft, and it appears neither in EIP-7773’s scheduled list nor in the Foundation’s announcement, so it activates neither today nor whenever mainnet follows. What Glamsterdam changes is how much a block can carry and how much of its validation can happen in parallel. The clock stays where it is.
The exit queue gets wider, and the safe resync window gets shorter
The change with the most direct line to a Canadian product is EIP-8061, which governs how fast staked ETH can leave the validator set.
| Today (Fulu) | Under Glamsterdam (Gloas) | |
|---|---|---|
| CHURN_LIMIT_QUOTIENT | 65,536 | 32,768 |
| Cap on ETH exiting per epoch | 256 ETH | none |
| Cap on ETH activating per epoch | 256 ETH | 256 ETH |
| Consolidation churn | shares the same budget | own quotient, 65,536 |
An epoch is 384 seconds, so 225 of them fit in a day. The present 256 ETH per-epoch cap therefore sets a hard ceiling of 57,600 ETH a day on exits, and that ceiling does not budge however much ETH is staked. Glamsterdam removes the cap outright and halves the divisor, making the exit limit total active stake divided by 32,768 per epoch. By our arithmetic on those two constants, the new rule is more permissive than the old cap at any total stake above 8,388,608 ETH, which is 256 multiplied by 32,768. EIP-8061’s own motivation refers to “roughly 32M ETH of 0x01 stake”, so the network sits far above that crossover.
The abstract puts the effect as “roughly doubl[es] the consolidation churn, as well as quadrupling the exit churn and restoring its proportionality to total stake”. The motivation names the event behind it: “the exit queue has stretched beyond forty days as a consequence of the mass exit of Kiln validators”.
There is a price, and the EIP states it. The parameters were chosen to balance “maintaining a sufficiently long weak subjectivity period (~7 days, roughly halving the current period)”, the window in which a node that has been offline can resync and work out which chain is real without trusting anyone’s word for it. Halving it is a deliberate security trade for faster exits. Separately, EIP-8045 stops validators the protocol has penalised for provable misbehaviour from proposing blocks.
Where the exit rules bind on a Canadian fund, and where they do not
The Purpose Ether ETF, one of the Canadian funds holding this exposure, has traded on the Toronto Stock Exchange since April 19 2021, and Purpose describes it as “the world’s first physically settled Ether ETF”. Purpose is the same issuer whose Bitcoin fund we track for the coins it actually holds reading by reading.
Two Canadian-dollar classes trade on the same fund, ETHH currency hedged and ETHH.B unhedged, both carrying a 1.00% management fee and a 1.42% MER. That is the same currency decision in two tickers, which we measured on Purpose’s Bitcoin pair last week in the hedged and unhedged return gap. NAV at October 5 2026 was $9.98 CAD for ETHH and $12.90 CAD for ETHH.B.
Purpose’s fund page, as it reads today, says the fund “stakes a portion of its Ether holdings to earn additional yield through Ethereum’s proof-of-stake mechanism”. Staked ETH sits inside a validator, and the exit churn governs how fast validators can be closed out entirely. Balance above a validator’s floor can be withdrawn without touching that queue, and a fund meeting ordinary redemptions runs an unstaked buffer, so the churn limit is not a Tuesday-morning constraint. It binds on a wind-down or a large repositioning of the staked book, which is the case EIP-8061 changes the arithmetic for.
At June 30 2026 the fund held 115,530 ETH at a fair value of $258.5 million CAD against an average cost of $378.2 million CAD, with net assets of $257.6 million CAD, and ETH has risen sharply in Canadian dollars since that balance sheet date, so the June position is not where the fund stands now. Its US dollar exposure that day was $258.6 million CAD gross, cut to $170.8 million CAD net of forward contracts, or 66.3% of net assets. A fund is one way to hold this exposure from a Canadian account; the operating companies in the sector are another, which our list of Canadian crypto stocks covers.
What the four upgrades before this one took to get from Sepolia to mainnet
No client team has committed to a mainnet date, so precedent is all there is. Each of the four network upgrades since the Merge forked Sepolia before it reached mainnet, and both timestamps are written into their own specifications.
| Upgrade | Sepolia fork | Mainnet activation | Days |
|---|---|---|---|
| Shapella | 2023-02-28 | 2023-04-12 | 43.8 |
| Dencun | 2024-01-30 | 2024-03-13 | 42.6 |
| Pectra | 2025-03-05 | 2025-05-07 | 63.1 |
| Fusaka | 2025-10-14 | 2025-12-03 | 50.6 |
The median gap is 47.2 days, the shortest 42.6 and the longest 63.1. Applied to today’s fork, that range maps to November 18 to December 8 2026, with the median landing on November 22.
Four observations are four, and nothing on that record obliges the fifth upgrade to behave like the first four. There is also a concrete brake on reading the window as a forecast. Each of the last three times, a further test network forked between Sepolia and mainnet, on dates recorded in the activation tables of the meta EIPs: Holesky on February 7 2024 for Dencun in EIP-7569, Hoodi on March 26 2025 for Pectra in EIP-7600, and Hoodi on October 28 2025 for Fusaka in EIP-7607. Shapella predates that practice and is recorded differently. Glamsterdam’s Hoodi row sits empty, so the first visible marker on the way to mainnet has not appeared. Take the window as a floor on how soon, not a forecast of when.
Four prior Sepolia forks, and no shape to what followed
Prices sat close to flat into the activation. ETH was at $2,713.04 USD, up 0.07% on the day, and $3,873.44 CAD, up 0.16%, on live bars rather than closes, pulled from Yahoo Finance at 7:05 a.m. ET. On daily closes, ETH is down 8.56% in 2026 against its 2025 year-end close of $2,967.04 USD.
Sepolia has forked ahead of four previous upgrades. Here is what ETH did after each, on Yahoo daily closes, with the event dates taken from the execution-layer fork timestamps written into Sepolia’s own config.
| Upgrade | Sepolia fork | ETH close that day | +30 days | +60 days | +90 days |
|---|---|---|---|---|---|
| Shapella / Shanghai | 2023-02-28 | $1,605.90 USD | +11.63% | +18.87% | +17.88% |
| Dencun / Cancun | 2024-01-30 | $2,344.49 USD | +42.54% | +49.62% | +37.15% |
| Pectra / Prague | 2025-03-05 | $2,241.98 USD | -19.03% | -19.33% | +15.67% |
| Fusaka / Osaka | 2025-10-14 | $4,125.41 USD | -21.64% | -24.45% | -25.04% |
| Median of the four | -3.70% | -0.23% | +16.77% | ||
| Positive | 2 of 4 | 2 of 4 | 3 of 4 |
Four events is not a base rate. Two are up and two are down at 30 days, and the four outcomes run from -21.64% to +42.54%. The smallest of them in absolute terms, +11.63%, is roughly the size of an ordinary month for ETH. The largest is nearly four times one. That is dispersion rather than a pattern.
ETH’s 30, 60 and 90 day returns after each of the four previous Sepolia forks.
The fair comparison is Ethereum’s own behaviour since the first of these four forks. Over the 1,286 overlapping 30-day windows since February 28 2023, ETH’s median move is +0.19%, positive 50.4% of the time, and its median absolute move is 11.40%. The fork median of -3.70% sits below that. At 60 days, -0.23% after forks against an ordinary +3.26% over 1,256 windows. At 90 days, +16.77% after forks against +0.69% over 1,226 windows, which cuts the other way. The comparison changes with the horizon you pick.
Disclaimer: The content on bestcanadianstocks.ca is for informational and entertainment purposes only and does not constitute financial advice. Past performance is not indicative of future results. Always consult a qualified financial advisor before making investment decisions. Fork timing, EIP contents and activation status come from first-hand protocol documents: the Ethereum Foundation’s Glamsterdam testnet announcement of September 28 2026, hardfork meta EIP 7773 and the individual EIPs, and the Sepolia and mainnet network configuration files published by the eth-clients project. The Sepolia activation second is fixed by the consensus-layer epoch 353,024 and the execution layer’s own amsterdamTime value, which agree. Ethereum prices are a Yahoo Finance snapshot taken at 7:05 a.m. ET on October 6 2026, live bars rather than closes. The post-fork return study and the Sepolia-to-mainnet gap study are our own arithmetic on those documents and on Yahoo daily closes. Purpose Ether ETF figures come from Purpose’s own published fund data and from the Purpose Funds 2026 interim financial statements to June 30 2026, Purpose Ether ETF section only.
