Ethereum's Glamsterdam hard fork, currently in Devnet-5 testing with a tentative June 2026 mainnet target, represents the network's largest execution-layer overhaul since the Merge. The upgrade centers on two headliner EIPs: EIP-7732, which enshrines proposer-builder separation into the consensus...
"We don't friggin need more copypasta EVM chains." — Vitalik Buterin, Ethereum Co-founder
Ethereum's Glamsterdam hard fork, currently in Devnet-5 testing with a tentative June 2026 mainnet target, represents the network's largest execution-layer overhaul since the Merge. The upgrade centers on two headliner EIPs: EIP-7732, which enshrines proposer-builder separation into the consensus layer, and EIP-7928, which introduces block-level access lists to enable parallel transaction processing. Combined with EIP-7904's gas repricing, the fork targets a 78.6% reduction in gas costs, a throughput increase from approximately 1,000 TPS to 10,000 TPS, and up to 70% less MEV extraction.
The stakes are structural. Three block builders currently produce over 90% of Ethereum blocks. The gas limit is set to rise from 60 million to 200 million per block. Layer-2 networks — Base, Arbitrum, and Optimism — now process nearly 90% of all L2 transactions and have captured users largely because L1 gas was prohibitively expensive. Glamsterdam's fee reduction could pull activity back to the base layer, altering the economic relationship between Ethereum and its rollup ecosystem.
Glamsterdam follows the Pectra and Fusaka upgrades delivered in 2025. Where Fusaka was primarily a data-layer upgrade addressing how Ethereum transmits data, Glamsterdam targets the execution layer — who produces blocks and how transactions are processed.
The Ethereum Foundation's DevOps team has tested three proposed EIPs on Devnet-4 and transitioned to Devnet-5 as of late March 2026. Public testnets with dual audit phases are planned for spring. The scope freeze occurred in late Q1 2026, locking the technical components. Core developers have stated that schedule adherence is secondary to correctness — a realistic slip to Q3 or Q4 2026 remains possible.
The upgrade includes roughly 50 non-headliner EIPs in addition to the two headliners. Coinbase's Base team publicly identified six priority proposals across three categories: blob scaling, gas repricing, and builder UX improvements, according to a post on the Base developer blog.
The centralization of block building is Ethereum's most acute structural problem. According to data from relay monitoring services, three builders — Beaver, Rsync, and Blocknative — produce over 90% of Ethereum's blocks. This concentration exists because new entrants must pay up to 1.4 ETH just to access private order flow, and providers impose reputation requirements evaluated on market share. The result is a self-reinforcing oligopoly.
EIP-7732 moves proposer-builder separation (PBS) from the current off-chain relay infrastructure — dominated by Flashbots' MEV-Boost system — directly into Ethereum's consensus layer. The mechanism replaces today's trusted relay model with an in-protocol commit-reveal flow.
Under the current system, validators outsource block construction to specialized builders via third-party relays. This introduces trust assumptions: validators trust relays to faithfully report block contents, and relays trust builders to deliver valid blocks. EIP-7732 eliminates both dependencies.
The expected impact is a reduction in MEV extraction of up to 70%, according to estimates cited by multiple Ethereum researchers. The mechanism standardizes MEV handoff rules at the protocol level, removing the asymmetric information advantages that currently benefit incumbents.
Flashbots has already begun its own decentralization effort. In late 2024, the organization launched BuilderNet, jointly operated with Beaverbuild and Nethermind, running on trusted execution environments (TEEs) and designed to distribute MEV. Flashbots ceased operating centralized block builders on Ethereum in December 2024. EIP-7732 would enshrine similar principles at the protocol level, making them enforceable rather than voluntary.
EIP-7928 introduces Block-Level Access Lists (BALs) — structured lists specifying which storage slots and accounts each transaction will access. BALs serve as an index that allows Ethereum clients to identify which transactions touch independent state, enabling those transactions to execute in parallel rather than sequentially.
Historical analysis of Ethereum transactions shows that 60-80% access disjoint storage slots. The remaining 20-40% can still benefit from parallelization through post-transaction state diffs included in the BALs.
The gas limit increase from 60 million to 200 million per block provides raw capacity. BALs provide the scheduling intelligence to use that capacity. Together, they target throughput of approximately 10,000 TPS, up from the current effective rate of roughly 1,000 TPS on the base layer.
BAL overhead is manageable: historical data indicates an average size of approximately 70 KiB per block. The access lists are constrained by the block gas limit rather than a fixed maximum, meaning the system scales with the network's existing resource pricing.
This is explicitly Phase 1 of parallel processing. Ethereum developers have stated that future upgrades will expand parallelism as client implementations mature and operational data accumulates.
EIP-7904 recalibrates gas costs for opcodes, precompiles, memory expansion, and data access operations. The repricing is anchored to empirical benchmarks from the Gas Cost Estimator project, which tested seven widely-used EVM implementations to measure actual computational effort.
Many of Ethereum's current gas prices were set years ago and no longer reflect execution costs on modern hardware. The EIP specifically targets 18 underperforming contracts benchmarked below 60 Mgas/s.
The repricing yields a 78.6% reduction in gas costs for both simple ETH transfers and complex smart contract interactions, according to the EIP specification. At current gas prices — approximately 0.085 gwei as of April 2, 2026 per Etherscan's gas tracker — a standard ETH transfer costs between $0.20 and $0.60. Post-Glamsterdam, these costs would fall further.
The methodology prioritizes computational complexity while explicitly excluding network-related costs such as state persistence, which are addressed separately in the upgrade's broader EIP package.
The fee reduction introduces a strategic tension between Ethereum's base layer and its Layer-2 ecosystem. The three dominant rollups — Base, Arbitrum, and Optimism — together process nearly 90% of all L2 transactions, with Base alone accounting for over 60%, according to a report from 21Shares.
Base was the only L2 to turn a profit in 2025, earning approximately $55 million, according to on-chain data. In February 2026, Base transitioned away from the OP Stack and ceased sharing sequencer revenue with the Optimism treasury.
The economic logic of L2 migration was straightforward: L1 gas was too expensive and MEV on L1 penalized retail traders. If Glamsterdam delivers a 78.6% gas reduction and 70% less MEV extraction, the cost differential between L1 and L2 narrows substantially. Some transaction volume could migrate back to the base layer, which carries a higher security guarantee than any rollup.
However, the relationship is not purely competitive. Both PeerDAS and EIP-7732 support increased blob capacity, providing L2s with additional data availability necessary to continue scaling. Coinbase's Base team endorsed this complementary dynamic in their Glamsterdam proposals blog post, identifying blob scaling as a priority category.
The net effect on L2 revenue remains uncertain. A cheaper L1 could erode L2 fee margins for simple transactions while L2s retain advantages for application-specific execution environments and user-experience customization. The market will price this in real time once Glamsterdam activates.
The 200 million gas limit imposes higher computational demands on validators. Professional solo stakers currently recommend 8-core CPUs, 32-64 GB of RAM, and 4-8 TB NVMe SSDs. Glamsterdam's increased throughput will push these requirements upward.
One mitigating factor: EIP-7732's ePBS extends Ethereum's proving window from 1-2 seconds to 6-9 seconds, making real-time proof generation feasible compared to the current seven-second average proving time that requires 12 GPUs. This relaxation benefits validators with consumer-grade hardware.
Longer-term, the introduction of zkAttesters — validators that verify zero-knowledge proofs rather than re-executing the full state — could reduce hardware requirements for home validators by eliminating the need for full execution-layer state storage. This feature is not part of Glamsterdam but appears on the Hegota upgrade roadmap (H2 2026).
The risk is real: if proof generation or block validation becomes dominated by well-capitalized operators, Ethereum's security model — which depends on distributed validation — weakens. Ethereum researchers have flagged this as a key design constraint for all throughput-increasing upgrades.
Glamsterdam is not an incremental upgrade. It restructures three foundational aspects of Ethereum — block production, transaction execution, and gas pricing — simultaneously. The on-time delivery of Pectra and Fusaka in 2025 provides some execution confidence, but Glamsterdam's scope is materially larger.
The economic implications extend beyond Ethereum's base layer. The $55 million annual revenue that Base generated as the only profitable L2 in 2025 depends partly on an expensive L1. A 78.6% gas reduction changes that calculus. Whether this consolidates Ethereum's position as the settlement layer or fragments its fee economy depends on how quickly the upgrade delivers and how L2s adapt their pricing models.
Vitalik Buterin declared 2026 "the year that we take back lost ground in terms of self-sovereignty and trustlessness." Glamsterdam is the mechanism. Whether it ships on time, and whether its throughput gains come at acceptable centralization cost, are the open questions.