Ethereum’s Glamsterdam hard fork went live on the Sepolia testnet today, October 6, at 13:53 UTC, hours after one of the network’s main validator clients shipped a fix for a bug found days earlier. The fork test-runs the biggest set of Ethereum changes since the Merge, from how blocks are built to how much work each block can carry.
Glamsterdam bundles two headliner changes. Enshrined PBS, or ePBS, takes the separation between block builders and proposing validators and moves it into the protocol itself, instead of running it through MEV-Boost, the out-of-protocol software most big validators use today. Block-Level Access Lists, BALs, require every block to publish up front which accounts and storage slots it touched.
The capacity number is the one operators are watching. Sepolia will target 200 million gas per block under the new rules, the second step of a doubling plan client teams agreed in September, up from 45 million at the start of 2026. The roadmap eventually lands at 1 billion, and client teams want the 200 million step exercised in public before any mainnet decision. Alongside ePBS and BALs, the fork also carries work on the peer-to-peer transport layer, raising the message size a node can push across the devp2p network.
Why the fix matters
The hours before the fork were not quiet. One of the main validator clients updated its software after a bug it found internally, fixing it shortly before the Sepolia test was due to raise per-block work to 200 million gas. Nodes that do not pick up the fix stop tracking the new blocks, and bigger Sepolia blocks mean more disk space per node. Testnet operators had hours to move, not weeks.
The risk of a stumble on Sepolia is real but contained. Commentary around the run-up has been blunt that a testnet cut-over does not stake real money, so the worst case is a stalled testnet and a fork that is delayed or reintroduced. The point of running a hard fork on Sepolia first is to find those stalls before they reach mainnet, where a dropped fork would freeze tens of billions of dollars in ETH for the duration.
The staking queue is a second problem
Separate from Glamsterdam, Ethereum investors who want to stake now wait more than two weeks before their coins start earning rewards, with roughly 1.5 million ETH, about $4.3 billion, sitting in line. The exit side of the queue is full too, around 458,000 ETH waiting to leave the staking system. That tells you large holders are adding new positions and old ones are leaving at the same time.
The churning queue exists by design. The protocol caps how fast validators can enter or leave so a runaway bug cannot drain the network in a day. It has only filled this badly a handful of times in Ethereum’s history, most notably in 2023 when the wait pushed past 40 days. The current one is a signal about demand, not a fault.
Why it matters for Glamsterdam is capacity. A 200 million gas target makes each block heavier for validators to build, verify and gossip. Client teams have been arguing for higher block size targets as the project scales L1 execution, and with more validators online, both the staking queue and the block weight move in the same direction. Nobody wants those two changes to collide untested.
What actually changes for users
For anyone who trades or holds ETH, Glamsterdam changes little today. Sepolia coin is worthless and the fork does not touch mainnet pricing. Rollup operators and L2 teams are the ones who care immediately, because Block-Level Access Lists give them cheaper proofs to verify and a cleaner data trail. Event-driven bots get a better signal for where MEV sits in a block, and staking pools get a protocol-level rule for what a builder can and cannot do. Individual users mostly notice it months later as slightly cheaper L2 fees, if it works.
How ePBS came to define this fork
The road from idea to fork was long. Ethereum researchers proposed enforcing the builder and proposer split inside the protocol years ago, and it sat in the plans while other upgrades shipped. Since the Merge in 2022, most big validators have used MEV-Boost, separate middleware that lets a proposer accept the best bid from professional block builders. Regulators and security researchers have called out the trust assumptions in that setup, since a validator technically signs one block while a second, more profitable block has already been handed to the network.
ePBS is how Ethereum decides to move that arrangement into the core client and stop depending on middleware that had grown to carry most big staking flows. It forces a proper adjustment of how much of the scaling story sits on the L1 versus rollups. There was a build-up phase where client teams went back and forth on scope before settling on ePBS plus BALs as the headline pair for the next hard fork.
BALs change security as much as speed. With access lists published with every block, nodes can double-check the state changes without replaying the whole block, and light clients can follow along cheaply. It also makes bugs visible, which cuts both ways. An attacker can read the access list as easily as a verifier, so privacy inside a block narrows. The gas target test on Sepolia puts BALs under maximum stress, because bigger blocks mean longer lists.
What comes next
Sepolia is the midpoint of the public test, with other Ethereum testnets carrying the same fork in the coming weeks. Mainnet timing is still open. Client teams have said the focus is on the testnets first, and the roadmap has not committed to a mainnet date.
The biggest thing to watch over the next two weeks is whether those 200 million gas Sepolia blocks hold under real load, and whether the validator clients stay in sync. Whatever happens on the testnet, the mainnet version of this code is months out, not days.
The fork has also become a reference point for other chains. Solana’s Votor consensus overhaul and Zcash’s 25-second blocks are both in flight, and several L2 operators say the cost of running their own chains is pushing smaller projects toward consolidation. Whether Glamsterdam makes Ethereum an easier network to run, or just a heavier one, Sepolia will start showing in the next two weeks.