Introduction

Glamsterdam is Ethereum’s next major protocol upgrade, and the name gives away how it’s built: “Amsterdam” for the execution-layer changes, “Gloas” for the consensus-layer ones.

Where Fusaka, which shipped to mainnet in December 2025, focused on foundational refinements, Glamsterdam is aimed squarely at scaling Ethereum’s base layer, reorganizing how the network processes transactions and manages its growing database.

The upgrade was originally targeted for June 2026, then pushed to the third quarter as the scope grew. In August, the target moved again to Q4 2026. The Ethereum Foundation launched Plataberget, the first public testnet built specifically for Glamsterdam, on 17 August, with the fork executed on the network three days later, on 20 August. Testing has continued past the originally planned schedule: an additional devnet, Devnet-11, started on 14 September, after the upgrade failed to run cleanly on a stable private devnet the first time around. Sepolia now has a confirmed date, 6 October 2026, set during the core developers’ call on 17 September, though the team has flagged a security concern specific to the test: free Sepolia test ether could let bad actors disrupt the rehearsal by repeatedly winning block-building auctions and withholding payloads. Hoodi follows, tentatively on 27 October. Mainnet still has no fixed date, though Q4 2026 remains the working target. The delay came with real progress attached: a 200 million gas limit floor for the post-upgrade network, up from roughly 60 million today.

What Glamsterdam actually changes

The upgrade centers on three goals: processing transactions in parallel instead of one by one, expanding how much data the network can safely handle, and pricing storage in a way that doesn’t quietly overload node hardware as the chain grows. Two proposals carry most of that weight.

• Enshrined Proposer-Builder Separation (ePBS) builds a separation that already exists informally, through third-party relay software like MEV-Boost, directly into the protocol. Today, the handoff between the validator that proposes a block and the builder that assembles it relies on off-protocol trust: a relay holds the block, shows the proposer only the bid, and only releases the contents once the proposer has signed a commitment it hasn’t actually seen. Enshrining that handoff removes the need for a trusted middleman: the builder locks in part of its balance on the beacon chain as a guarantee, a new validator committee (the Payload Timeliness Committee) checks whether the payload was revealed on time, and the protocol settles payment automatically. Relays like MEV-Boost don’t disappear, they become optional rather than load-bearing, still useful for features the core protocol doesn’t yet support natively. In the process, the window validators have to propagate data across the network stretches from about 2 seconds to roughly 9, which is what makes handling larger blocks and more blobs viable without straining the network.

• Block-Level Access Lists (BALs) solve a different bottleneck. Ethereum currently has to process transactions sequentially, because it doesn’t know in advance which parts of the database a transaction will touch. BALs attach that information upfront, so validators can identify which transactions don’t overlap and process them at the same time. As a side effect, it also lets nodes sync by copying final results instead of replaying every transaction from scratch.

• A smaller addition worth flagging, especially for institutions: EIP-7708 makes native ETH transfers emit a log, the same way ERC-20 tokens already do. Today, tracking raw ETH movements, especially transfers that happen inside a contract call, means tracing every transaction, since a plain ETH transfer leaves nothing for a standard log query to find. Once EIP-7708 is live, ETH becomes trackable through the same simple log subscriptions wallets, exchanges and indexers already use for tokens, simplifying reconciliation and reporting without changing how ETH itself works.

Why the Gas limit number matters

A 200 million gas limit sounds abstract until you compare it to where the network sits now. It’s roughly 3.3 times current capacity, and it’s not a value Glamsterdam forces on the network directly. Validators still raise the limit gradually through gas-vote signaling, the way they always have. What Glamsterdam does is remove the technical reasons that limit couldn’t safely go higher.

That headroom needs a counterweight, which is where EIP-8037 comes in. Adding accounts or deploying contracts creates permanent data every node has to store indefinitely, and today’s fees don’t really reflect that long-term cost. EIP-8037 introduces a fixed cost per state byte, targeting a state growth rate of around 120 GiB a year, specifically so that raising the gas limit doesn’t quietly price standard hardware out of running a node. Home staking staying accessible is very much the point here, not a side effect.

What changes operationally

Glamsterdam isn’t a passive upgrade for anyone running infrastructure. Because ePBS changes how blocks get built and validated, both execution and consensus clients need updated versions, and node operators need to track client releases to stay in sync post-fork.

A smaller but practical fix worth knowing: today, exits and validator activations share a single capped limit of 256 ETH per epoch, which meant queues could stretch past 40 days during periods of heavy exit demand, as happened during a security incident late in 2025. EIP-8061 reorganizes this into separate lanes: exits are no longer capped at a fixed number and instead scale with total ETH staked, pushing the limit to roughly 1,258 ETH per epoch at current stake levels, a near 5x increase, while consolidations get their own dedicated capacity. A competing proposal, EIP-8080, would have let standard exits borrow spare capacity from the consolidation queue instead; it was considered through the spring but was ultimately declined in favor of EIP-8061’s broader fix. For anyone planning withdrawal timing at scale, the practical upshot is a shorter, more predictable exit queue.