Ethereum has scheduled the first public test of its Glamsterdam upgrade for October 6, when the fork activates on the Sepolia testnet at 13:53 UTC. Testnet operators have been asked to upgrade their clients by October 5. A mainnet launch is still expected in the fourth quarter, though no official date has been confirmed.
The checkpoint gives the scaling roadmap a hard date after months of devnet work. Ethereum Foundation researcher Toni Wahrstätter described the upgrade as close to its final development stage, with client teams now moving from private testing to a public environment where validators and infrastructure providers can stress the code before real value sits on the line.
What Glamsterdam changes
The centerpiece is EIP-7928, block-level access lists. Today a block executor touches state in whatever order transactions demand, which makes parallel execution hard to prove and slow to verify. A block-level access list records every account and storage slot a block touches, so nodes can fetch that state ahead of execution and run non-conflicting transactions at the same time. The same structure speeds up sync: new nodes can pull a compact view of what each block touched instead of replaying everything from scratch, an approach tied to the snap v2 sync protocol.
Glamsterdam also carries ePBS, enshrined proposer-builder separation, which moves block construction and block proposal into the protocol itself rather than leaving it to off-chain agreements between validators and builders. The 2027 Hegota fork, which Buterin described in his long-form vision last week, builds on both pieces.
The gas limit climb
Sepolia will signal a 200 million gas limit, up from roughly 60 million on mainnet today. The rise is not a switch that flips at fork time. Testnet validators raise it gradually through block-by-block voting, which lets capacity grow while client teams watch for stalled blocks or memory strain. If the pattern holds, the same mechanism becomes the template for raising the mainnet limit after the fork ships.
Higher gas limits only help if execution keeps up, which is the point of the access lists. Parallel execution and faster sync are what let node operators absorb a bigger block without paying for it in hardware, a trade-off that has defined every recent argument about Ethereum scaling.
Why block-level access lists matter
The proposal changes what a node knows before it executes. Right now a validator learns which state a transaction needs only while running it, which serializes work that could in principle run side by side. With an access list agreed at block level, execution can be planned: conflicting transactions are ordered, independent ones are dispatched in parallel, and the result is verifiable because the list itself is part of consensus data.
Client teams have been prototyping against this model on devnets for months. The Sepolia fork is the first time the machinery runs where anyone can connect a node, submit transactions and watch how reth, geth, nethermind and the other clients behave under a rising gas limit. Bugs found here are cheap; the same bugs on mainnet are not.
There is a cost on the other side of the ledger. Recording a full access list makes blocks bigger, which means more bandwidth and disk for node operators. The design tries to offset that with faster sync and cheaper verification, but the balance is exactly what Sepolia is meant to test under load. Skeptics within the client community have argued the overhead lands hardest on home stakers, the group the roadmap claims to protect.
What it means for users and builders
For rollups and L2 operators, block-level access lists shorten the wait for state proofs and make data availability cheaper to verify. For ordinary users the visible effect arrives later: cheaper transactions when the mainnet gas limit rises, and faster node sync for anyone running infrastructure. None of it is live on mainnet yet, and the Q4 window depends on Sepolia staying clean through October.
ETH traded near $2,650 on Monday, down from higher levels last week, as the broader market slipped on oil-driven risk aversion. Spot ether ETFs took in $690 million last week after outflows the week before, so the upgrade news lands while institutional demand has already turned back positive.
The schedule also lands in a crowded stretch for the protocol. Client teams are juggling Glamsterdam work alongside the post-quantum roadmap that Buterin sketched last week. Developers have said the Hegota headline features are already running on devnets, which keeps the 2027 target plausible if Glamsterdam does not slip.
What happens after October 6
The Sepolia run is the first of two public checkpoints. If it holds, expect a Holesky or mainnet shadow fork next, then a dated mainnet announcement. If client bugs surface, the Q4 window slips, as it has before. Either way the activation gives the community a concrete test of whether the access-list design survives contact with a live network, and it starts the clock on the gas-limit debate for mainnet.
Traders will watch two numbers during the testnet run: the realized gas limit as validators vote it upward, and block propagation times as blocks grow. Both feed directly into the mainnet conversation, where a 200 million limit would roughly triple current capacity. The Foundation has not committed to a mainnet figure, saying only that the testnet signal is a stress test rather than a promise.
For anyone running infrastructure, the practical advice from client teams is simple: upgrade to the Glamsterdam-ready releases before October 5, watch the client dashboards during the fork hour, and report anomalies to the testing channel. The fork itself is expected to be uneventful if clients ship clean releases, and the interesting data will come in the following days as the gas limit climbs.