Ethereum's Glamsterdam hard fork — the network's largest protocol overhaul since the September 2022 Merge — has locked in ten EIPs and entered final devnet testing ahead of a projected November 4, 2026 mainnet activation. The two headline proposals, EIP-7732 (Enshrined Proposer-Builder Separation...
"This is probably the largest fork we've had since the Merge. It will change a lot of assumptions about Ethereum and set us up for much more scaling in the future." — Parithosh Jayanthi, Core Developer, Ethereum Foundation
Ethereum's Glamsterdam hard fork — the network's largest protocol overhaul since the September 2022 Merge — has locked in ten EIPs and entered final devnet testing ahead of a projected November 4, 2026 mainnet activation. The two headline proposals, EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists), together restructure how blocks are built and executed on Ethereum's base layer.
The upgrade targets a gas limit increase from the current 60 million to 200 million per block, enabling approximately 10,000 transactions per second on L1 — a tenfold increase from today's effective throughput. A companion gas repricing package is projected to reduce average L1 transaction fees by approximately 78.6%. Sepolia testnet fork is scheduled for September 21, 2026, with Hoodi testnet following on October 5.
The stakes are measurable. As of August 2026, 41.41 million ETH (33.98% of circulating supply) sits in staking contracts across roughly 1.24 million validators. Over 80-90% of Ethereum blocks are currently built through external relay networks like MEV-Boost. Glamsterdam aims to reduce dependence on these trusted intermediaries by moving proposer-builder separation into the protocol itself.
Glamsterdam pairs an execution-layer upgrade (codenamed "Amsterdam") with a consensus-layer upgrade (codenamed "Gloas") into a single coordinated hard fork. It follows Fusaka, which shipped in December 2025, and precedes Hegotá, targeted for H1 2027.
The upgrade ships ten EIPs total, organized into three functional categories:
| Category | EIPs | Primary Effect | |----------|------|----------------| | Block production | EIP-7732 | On-chain proposer-builder separation | | Parallel execution | EIP-7928 | Block-level access lists enabling parallel validation | | Gas repricing | EIP-7976, EIP-7981, EIP-8037, EIP-7778, EIP-7954, EIP-8024, EIP-7843, EIP-7708 | Fee structure adjustments, state cost increases, new opcodes |
The combined effect: higher throughput, cheaper compute operations, more expensive state creation, and reduced reliance on off-protocol block-building infrastructure.
Today, Ethereum's block production pipeline works through an off-protocol system. Validators outsource block construction to specialized builders through relay networks such as MEV-Boost, operated by Flashbots and others. According to relay data, 80-90% of Ethereum blocks are built this way. The system functions, but it requires validators to trust third-party relays — entities that are not subject to protocol-level accountability.
EIP-7732 formalizes this handoff inside the consensus layer. Under ePBS:
The existing MEV-Boost relay system will continue to function. The new ePBS market is strictly opt-in — there is no forced migration. However, because on-chain builders operate without relay trust assumptions, Ethereum Foundation researchers project up to 70% less MEV extraction leakage compared to the current system.
A critical architectural change: EIP-7732 extends the block propagation window from approximately 2 seconds to approximately 9 seconds. This longer window enables future blob capacity increases — potentially reaching 72 or more blobs per block — clearing a bottleneck that constrained Fusaka's PeerDAS scaling gains.
According to Ethereum Foundation core developer Parithosh Jayanthi, Glamsterdam will "majorly change the cost of actions on Ethereum. High-level compute gets cheaper and state gets more expensive."
EIP-7928, authored by Ethereum researcher Toni Wahrstätter, addresses a structural limitation: Ethereum currently processes transactions sequentially, one at a time, regardless of available hardware parallelism. As Wahrstätter stated, "Ethereum has effectively been running in single-lane mode: one long queue of transactions, executed strictly in order. Even with multi-core CPUs, validation stays sequential."
Block-Level Access Lists (BALs) attach a data structure to every block that maps every account and storage slot the block touches, along with post-execution state values. This mapping enables:
BALs are the prerequisite for the gas limit increase from 60 million to 200 million. Without parallel execution, raising the gas limit would simply increase the time needed to validate each block, defeating the purpose. With BALs, nodes can exploit multi-core hardware to validate larger blocks within the same time constraints.
The target throughput at 200 million gas with BALs: approximately 10,000 transactions per second on L1, compared to roughly 1,000 TPS today.
Beyond the two headline proposals, Glamsterdam ships eight EIPs that adjust gas economics and add execution-layer features:
Fee restructuring:
Execution-layer features:
The net projected effect: a 78.6% reduction in average L1 fees, driven primarily by the capacity increase rather than fee subsidies. As Jayanthi noted, the repricing deliberately makes state creation more expensive while making computation cheaper — a structural rebalancing rather than a blanket reduction.
| Milestone | Date | Status | |-----------|------|--------| | Soldøgn interop devnet | Concluded May 2, 2026 | Complete | | Final devnet (devnet-7) | June 2026 | Complete; stable under elevated transaction loads | | Plataberget public testnet | July-August 2026 | Active | | Sepolia testnet fork | September 21, 2026 | Scheduled | | Hoodi testnet fork | October 5, 2026 | Scheduled | | Mainnet activation | November 4, 2026 | Projected (unconfirmed) |
The Soldøgn interop devnet demonstrated multi-client ePBS execution across nearly all client implementations. Devnet-7, according to Ethereum Foundation communications, "held up well over the weekend with increased transaction loads." Developers tested builder behavior under pressure and iterated on gas semantics.
The original target was H1 2026. The Ethereum Foundation pushed to Q3, then to Q4, citing ePBS implementation complexity as the primary bottleneck. Cross-client implementation parity — ensuring Geth, Nethermind, Besu, Erigon, and Reth all handle the new block production pipeline identically — remains the gating factor for a confirmed mainnet date.
For validators and stakers: Both consensus-layer and execution-layer clients require updates before mainnet activation. The ePBS mechanism changes the validator's role — proposers no longer assemble execution payloads, which may reduce computational requirements for solo stakers while introducing new collateral dynamics for builders. Institutional stakers gain faster exit processing, reducing the operational cost of time-to-liquidity.
For L2 rollups: The extended 9-second propagation window unblocks future blob scaling beyond Fusaka's PeerDAS limits. L2 transaction costs, already reduced 90-99% by EIP-4844 (from $0.50-$5.00 to $0.001-$0.05), may see further compression as blob capacity increases in subsequent forks enabled by Glamsterdam's architecture.
For L1 applications: A 78.6% projected fee reduction and 10x throughput increase could make certain applications economically viable on L1 that currently require L2 deployment. DeFi protocols, NFT marketplaces, and high-frequency settlement systems stand to benefit from cheaper compute, though higher state creation costs will affect storage-heavy applications.
For the MEV supply chain: ePBS creates a permissionless alternative to the current relay-dependent system. Flashbots, BloXroute, Aestus, and other relay operators face structural competition from on-chain builders who operate without intermediary trust requirements. The relay market will not disappear immediately — the opt-in nature of ePBS ensures coexistence — but the economic rationale for relays weakens over time.
Centralization of block building. Vitalik Buterin warned in a March 2026 blog post that ePBS alone does not resolve the risk that a small number of builders could dominate and censor transactions. He proposed additional measures — including FOCIL (Forced Inclusion Lists), where 16 random attesters require transaction inclusion, and encrypted mempools that hide transaction data until block finalization. These are post-Glamsterdam proposals, not included in the current upgrade scope.
Hardware requirements. A 200 million gas limit with parallel execution demands more from node hardware. If block capacity increases but hardware costs rise significantly, the network may become increasingly reliant on institutional operators. The decentralization impact is unquantified.
Untested at mainnet scale. The interplay between ePBS and BALs has been tested on devnets but not under mainnet conditions — real MEV competition, real economic stakes, real validator diversity. The Base engineering team publicly warned that adding FOCIL alongside ePBS could delay the upgrade beyond 2026; that proposal was ultimately excluded.
Builder collateral dynamics. On-chain builders under ePBS must post ETH collateral. The collateral requirements and slashing conditions introduce new capital efficiency considerations that may affect builder participation rates and market structure.
Glamsterdam represents Ethereum's most ambitious base-layer restructuring since the Merge — a simultaneous overhaul of block production economics, execution parallelism, and gas pricing. The upgrade addresses quantifiable infrastructure weaknesses: relay dependence affecting 80-90% of blocks, sequential execution limiting throughput to ~1,000 TPS, and gas economics that underprice state growth.
Whether the November 4 mainnet target holds depends on cross-client testing outcomes over the next 11 weeks. The Ethereum Foundation has not formally confirmed the date. What is confirmed: devnets are stable, the EIP scope is locked, and testnet forks have scheduled dates. The engineering is converging. The open questions are about edge cases at scale — the kind that only emerge under mainnet conditions with real economic incentives.
For the Ethereum ecosystem's $228 billion market capitalization and 41.41 million staked ETH, the practical question is not whether Glamsterdam ships, but whether the upgrade achieves its capacity targets without the centralization side effects that Buterin has explicitly flagged as unresolved.