Ethereum’s biggest technical overhaul in years just cleared its first live rehearsal — and the countdown to the second one starts now. Glamsterdam passed its Sepolia testnet trial on October 6, developers have set October 26 for the final public testnet, and early December dates are already being floated for mainnet. Here is what is actually changing under the hood, in plain English.
By Keisha Williams | October 14, 2026
The Hook: Ethereum Is Rewiring Its Engine Room
Glamsterdam is Ethereum’s next major network upgrade, following Fusaka, which activated in December 2025. It went live on the Sepolia testnet on October 6, 2026 — an on-time landing that independent Ethereum researcher Christine D. Kim reported “went smoothly.” The next milestone is already scheduled: the Hoodi testnet on October 26, the second and final public rehearsal before the real thing. No mainnet date has been formally agreed, but a draft scheduling proposal on GitHub floats December 1, December 8 and December 9 as candidate slots.
While ETH trades near 2,416 USD — down amid a broader market slip — the upgrade itself is what long-term holders should understand. Where Ethereum’s previous two upgrades were mostly about staking and data capacity for layer-2 networks, Glamsterdam is about execution: the part of the system that actually runs your transactions. Think of the last upgrades as widening the pipes; this one rebuilds the engine room.
The Evidence: 18 Changes, Three Big Ideas
The upgrade’s scope sits in tracking proposal EIP-7773, which schedules 18 core changes. Most of them cluster around three big ideas — and each one matters for what Ethereum will cost and handle in the future:
- Enshrined proposer-builder separation (ePBS, EIP-7732) — moves the block-building handoff into the protocol itself. Today, most blocks are assembled by specialized outside builders and passed through middlemen called relays. If a relay fails or censors, Ethereum’s own rules offer no recourse. ePBS makes builders protocol participants with their own stake, removing the need to trust the middlemen.
- Block-level access lists (EIP-7928) — today, Ethereum largely processes a block’s transactions one after another, because it cannot know in advance which accounts each transaction will touch. Access lists give every block a map of exactly that, letting computers process non-overlapping transactions in parallel — like opening more checkout lanes at a store because you finally know who is buying what.
- Gas repricing (EIP-8037 and EIP-8038) — the first overhaul of state costs since 2021. Creating new accounts and storage slots becomes several times more expensive, while everyday transfers are preserved. The aim: stop the blockchain’s memory (its “state”) from bloating as capacity grows.
The payoff target is a 200 million gas limit floor after Glamsterdam — more than three times today’s 60 million default. The Ethereum Foundation says the new pricing supports roughly a 3x increase in base throughput, and ethereum.org says the upgrade “will most likely reduce fees for everyday users,” though that depends on how much new activity floods in to fill the extra room.
The Core Conflict: Bigger, Faster — but Not Free
Glamsterdam is not a pure giveaway, and that is where the debate lives. Staking economics shift in both directions. Validators gain faster exits — under EIP-8061, exit capacity roughly quadruples at current staking levels — and a protocol-enforced way to sell block space. But the safety window for nodes that go offline shrinks: ethereum.org says the so-called weak subjectivity period falls from about 15.7 days to about 7 days, meaning a node that goes offline for a long stretch needs a trusted checkpoint to rejoin safely.
Early Sepolia data also shows a thin builder market. According to a summary of the October 8 execution-layer call, only four or five builders were active on the testnet, with major builders absent. ePBS only proves its value when real builders genuinely compete — which makes the October 26 Hoodi run the more telling test.
And the cost redistribution is real: deploying large contracts becomes roughly eight times more expensive under the new pricing, while wallets and node operators must update their gas estimates or users will see failed transactions. A small set of older contracts with hardcoded gas assumptions may degrade, though the Foundation says the large majority of smart contracts are unaffected.
What This Means for Your Wallet
For ordinary holders, the most important guidance is the simplest: you do not need to do anything. The Foundation has explicitly said mainnet users need not act, and ethereum.org warns that anyone telling you to “upgrade” your ETH is running a scam. No migration, no new token, no claiming step — ever, for any legitimate Ethereum upgrade.
The benefits arrive on their own if all goes well: more block space, likely lower fees at the base layer, and a network that can safely carry far more activity — which matters whether you transact on Ethereum directly or use a layer-2 built on top of it. The extra capacity also means more room for the data that layer-2 networks depend on, so cheaper mainnet plumbing tends to ripple outward to the whole ecosystem.
What to watch on the calendar: client releases for Hoodi are expected by October 16, the Hoodi activation lands October 26, and a mainnet date should firm up shortly after if it goes cleanly. Those dates are when short-term ETH traders typically position around upgrade news — another reason long-term holders should know the schedule and then ignore the noise around it.
The Verdict
Glamsterdam is a statement of direction. For years, Ethereum pushed most scaling work onto layer-2 networks and kept its base layer conservative. This upgrade visibly changes that emphasis: the base chain itself is being rebuilt to process far more activity, price its resources honestly, and cut out trusted middlemen. The Sepolia rehearsal went smoothly, the December mainnet window is on the table, and the pieces — bigger blocks, parallel processing, protocol-enforced block building — fit together as a coherent engineering plan rather than a marketing one. Upgrades of this size always carry execution risk until the final testnet passes. But if Hoodi goes as cleanly as Sepolia did, Ethereum ends 2026 with a fundamentally more capable foundation — and everyday users get the best kind of upgrade: one they never have to think about.
The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.
7732 finally deleting the relay middleman and eth sits flat at 2416. the market stopped pricing upgrades after the merge, cheaper entry for the rest of us
eth flat at 2416 through the biggest protocol change since the merge is the clearest tell the market stopped pricing upgrades. window if hoodi behaves
flat price while shipping the relay fix is the best deal in crypto right now. upgrades stopped being catalysts after the merge but the compounding never stopped
Everyone arguing december 8 vs january while the real story is 7732 finally shipping ePBS. Relays losing block control matters far more than the fork date.
epbs finally enshrined after years of relays skimming the middle. december mainnet would be a great late birthday present
Careful with the december 1 vs 8 vs 9 talk, those are draft slots on the github proposal, not agreed dates. Hoodi on the 26th going clean matters more.
ePBS finally shipping is huge. relays censoring blocks has been an open wound for years and EIP-7732 just deletes the middleman from the protocol
agreed, but ETH sitting at 2416 during the biggest execution overhaul since the merge tells you the market does not price tech anymore
7732 keeps relays alive, builders just stop needing them to touch every single block. small difference but it matters for how censorship resistance actually plays out
With builders not needing relays to touch every block, censorship resistance finally lives in the protocol itself. That is the quiet win here.
eth at 2,416 while the biggest execution rework in years ships. classic
Hoodi on October 26 is the one that actually matters. Sepolia is the easy testnet, if Hoodi lands clean the December 1 slot gets real
hoodi going clean is priced at zero with eth at 2416. one missed epoch tho and the december drafts turn into january real fast
w take. one rough Hoodi run and that december draft proposal gets quietly shuffled to january
hoodi has burned schedules before tho. one missed epoch on the 26th and that december 1 slot is gone
the draft slots on that proposal already got reshuffled once this week. january is sitting there pre written, they just havent clicked submit
reshuffled draft slots once in a week is normal github churn tbh. fusaka slipped for months, this one actually has testnet momentum behind it
january is sitting there pre written made me laugh because its true. one slip on hoodi and the dec drafts vanish overnight
sepolia went smoothly, hoodi next. and then we wait for mainnet to get spicy
Fusaka landed on time and Glamsterdam is smaller scope than the merge was. i get the skepticism but december 8 is more likely than another slip
dec 8 over another slip feels right, but only if hoodi behaves on the 26th. sepolia was the easy exam, that one decides the december slot
fusaka landed but only after slipping twice. the momentum is real until hoodi on the 26th decides it isnt
Sepolia landing cleanly on the first try is the part nobody expected. October 26 on Hoodi is when I will actually start planning around a December fork date.
dec 1, 8 or 9 all listed as draft slots tells you even the devs wont commit. sepolia was the easy one, hoodi is where schedules go to die
fusaka slipped twice and still landed. glamsterdam has half the scope and sepolia went clean first try, hoodi on the 26th is the only real gate left
hoodi is also where the mev crowd stress tests epbs with real block flow for the first time. sepolia traffic is playground volume, mainnet builders are a different animal
genuine question, do exchanges need to change anything for epbs or is it fully protocol side. last thing anyone needs is another wave of deposit pauses over a fork