Ethereum's Glamsterdam hard fork, tentatively targeted for June 2026, represents the protocol's most structurally significant upgrade since Pectra. Two headline EIPs — EIP-7732 (enshrined proposer-builder separation) and EIP-7928 (block-level access lists) — aim to dismantle the off-chain relay i...
"ePBS only prevents builder centralization from spilling over into the staking layer, but the issue of builder centralization itself remains." — Vitalik Buterin, Ethereum Co-Founder
Ethereum's Glamsterdam hard fork, tentatively targeted for June 2026, represents the protocol's most structurally significant upgrade since Pectra. Two headline EIPs — EIP-7732 (enshrined proposer-builder separation) and EIP-7928 (block-level access lists) — aim to dismantle the off-chain relay infrastructure that currently governs 90% of block production and enable parallel transaction execution for the first time.
According to the Ethereum Foundation's Checkpoint #9, published April 10, 2026, Devnet-4 testing is complete and Devnet-5 is underway, with the first generalized Glamsterdam devnet expected within days. A third component — a comprehensive gas repricing package anchored by EIP-7904 and EIP-8037 — recalibrates state access costs that have remained misaligned since Ethereum's early years. Together, the three pillars target the protocol's most persistent structural weaknesses: MEV-driven centralization, sequential execution bottlenecks, and state bloat running at approximately 1.2 TB for full nodes.
This report examines the technical architecture of each component, quantifies the economic problem each addresses, and assesses the operational risks of a fork that touches consensus, execution, and gas economics simultaneously.
Approximately 90% of Ethereum blocks are currently built via MEV-Boost, an off-chain relay system developed by Flashbots. Two entities build roughly 90% of all blocks, according to Flashbots' own disclosures — a concentration that has tripled since 2022. The problem compounds: major orderflow providers have entered exclusive deals with block builders to internalize MEV, creating a self-reinforcing oligopoly.
The numbers illustrate the economic weight at stake. Cumulative MEV extraction on Ethereum surpassed $1.8 billion by mid-2025. Between December 2025 and January 2026, searchers captured approximately $24 million in MEV profit over a 30-day window. According to a July 2025 ESMA TRV Risk Analysis, sandwich attacks accounted for 51.56% of Ethereum's $561.9 million in MEV transaction volume during 2025, totaling roughly $289.8 million. A single operator, known as jaredfromsubway.eth, was responsible for approximately 70% of all Ethereum sandwich attacks in 2025.
The current relay landscape reflects this fragmentation. As of late October 2025, Ultra Sound relay handled 32.3% of MEV-Boost payloads, Titan Relay processed 24.75%, the two bloXroute relays combined for approximately 26%, and Flashbots' own relay had fallen to 3.44%. Ethereum's block production pipeline depends entirely on these off-chain intermediaries — entities that operate outside protocol consensus, with no on-chain accountability.
EIP-7732 moves proposer-builder separation from the relay layer into Ethereum's consensus protocol. The mechanism splits block production into two distinct roles enforced on-chain:
Builders assemble execution payloads — the ordered list of transactions that determines MEV — and cryptographically commit to their contents without revealing them. Proposers select the highest-paying builder commitment and include it in the consensus block. Transactions are revealed only after the block is finalized, eliminating the proposer's ability to inspect, reorder, or censor the payload.
The design introduces explicit deadlines, payload commitments, and fallback behavior. If a builder fails to deliver after committing, the protocol handles non-delivery without halting liveness — a significant improvement over the current system where relay failures can cause missed slots.
The structural change is significant: the protocol now handles "partial blocks" and two-party coordination within consensus. According to Checkpoint #9, this touches "practically everything" in the stack, as every component must reason about the possibility of disagreement between proposer and builder.
Buterin has been explicit about ePBS's limitations. The upgrade prevents builder centralization from contaminating the validator set, but it does not solve builder centralization itself. "Toxic MEV (sandwich attacks, front-running) may merely persist elsewhere," he stated. The Foundation's longer-term roadmap includes FOCIL (Forced Inclusion Lists) and encrypted mempools as subsequent measures — both assigned to the Hegota upgrade.
Ethereum currently executes transactions sequentially. EIP-7928 introduces Block-Level Access Lists (BALs), an enforced per-block data structure that records every account and storage location accessed during execution, along with post-execution state diffs (storage keys, balances, nonces, code).
The practical effect is parallelization across four dimensions:
The overhead is modest. BALs add approximately 35 KiB per block at the current 36M gas limit. Worst-case sizes remain below existing calldata worst cases.
The EIP has received broad support from core developers. Multiple client teams are prototyping implementations, and BAL-specific devnets (bals-devnet-2) are in active stress testing. Toni Wahrstätter, one of the EIP's authors, noted that the proposal "received much love from core devs recently."
For validators and node operators, BALs represent a shift in hardware demands. Recommended specifications for 2026 already call for 4-8 TB NVMe SSDs with read speeds exceeding 7 GB/s, 32-64 GB RAM, and 300-500 Mbps minimum bandwidth. Parallel execution could partially offset these requirements by improving throughput per unit of hardware.
The third pillar — less visible but arguably most consequential for daily users — is a comprehensive gas repricing package. EIP-7904 recalibrates gas costs to reflect actual computational resource consumption on modern hardware. Many current gas prices were set years ago and no longer reflect execution reality.
The recalibration yields a projected 78.6% reduction in gas costs for both simple ETH transfers and complex smart contract interactions, according to Datawallet analysis.
EIP-8037, a complementary proposal, moves in the opposite direction for state creation specifically. It raises gas costs for deploying new contracts and opening new storage slots — operations that permanently expand Ethereum's state footprint. Until now, state creation consumed the same gas type as computation, meaning any increase in compute throughput automatically increased state bloat.
The fifth Glamsterdam repricings call on April 1, 2026, focused primarily on EIP-8037. Developers are holding broader repricing decisions until updated benchmarking data is available, indicating the gas cost schedule remains in flux.
This decoupling of state creation from execution gas is a structural prerequisite for the Foundation's stated goal of pushing gas limits "toward and beyond 100M" without proportionally expanding the 1.2 TB state burden that full nodes currently bear. Archive nodes already require approximately 15 TB of storage.
As of the Ethereum Foundation's Checkpoint #9 (April 10, 2026), the development status is as follows:
| Component | Status | |-----------|--------| | EIP-7732 (ePBS) | epbs-devnet-0 in progress | | EIP-7928 (BALs) | bals-devnet-2 stress testing | | Gas repricing | Benchmarking ongoing | | Generalized devnet | Expected "next week" per April 10 post | | Public testnet | Planned for spring 2026 | | Dual audit rounds | Pre-mainnet requirement | | Mainnet target | June 2026 (aspirational) |
Developers have consistently described the June target as aspirational. The ePBS component is the primary risk to the timeline: it introduces two-party coordination at the consensus layer, requiring every client to handle partial block states and builder failure modes. The Checkpoint #9 post noted that the first generalized devnet depends on "the current ePBS devnet [being] stabilized."
The Ethereum Foundation's February 2026 Protocol Priorities Update organized all 2026 work into three tracks: Scale (led by Ansgar Dietrichs, Marius van der Wijden, Raúl Kripalani), Improve UX (focused on native account abstraction and cross-L2 interoperability), and Harden the L1 (led by Fredrik Svantes, tied to the Trillion Dollar Security Initiative). Glamsterdam addresses primarily the Scale track, with security hardening deferred to Hegota.
Over 1.2 million validators are active across 80+ countries. The validator set expanded significantly into West Africa, Southeast Asia, South America, and Eastern Europe, though the U.S. and Germany remain dominant. EIP-7251, activated with Pectra in May 2025, raised the maximum validator balance from 32 ETH to 2,048 ETH, consolidating the validator set and reducing peer-to-peer network overhead.
Glamsterdam is the first of two hard forks planned for 2026. Hegota, targeting H2 2026, carries its own headline features:
Verkle Trees: A data structure replacement that could reduce node storage requirements by approximately 90% and enable fully stateless clients — nodes that can validate blocks without storing the full state. This addresses the fundamental constraint that the current 1.2 TB state size imposes on decentralization.
FOCIL (Forced Inclusion Lists): A censorship-resistance mechanism that complements ePBS by allowing proposers to force specific transactions into blocks, preventing builders from systematically excluding certain addresses or transaction types.
Account Abstraction: Confirmed as a "minor feature" of Hegota's scope. Native account abstraction enables smart contract wallets as first-class citizens, removing reliance on ECDSA-based authentication — a prerequisite for post-quantum readiness.
Non-headlining EIP proposals for Hegota opened for submission on April 9, 2026.
Glamsterdam is a structural fork, not a feature upgrade. It rewrites how blocks are produced at the consensus layer, how transactions are executed at the computation layer, and how gas costs are denominated across the protocol. The three components are interdependent: ePBS changes who builds blocks; BALs change how those blocks are processed; gas repricing changes what it costs to include transactions in them.
The risk profile is proportional. Touching consensus, execution, and economics in a single fork creates surface area for unexpected interactions. Developers have acknowledged this explicitly by conditioning the timeline on ePBS devnet stabilization and requiring dual audit rounds before any mainnet deployment.
If Glamsterdam ships on schedule, Ethereum will have eliminated its dependency on off-chain relay infrastructure for block production, enabled parallel transaction execution, and recalibrated a gas cost schedule that has been misaligned for years. If it slips — which developers have signaled is possible — the delay reflects the complexity of changing three foundational systems simultaneously.
The upgrade does not solve builder centralization outright. That falls to FOCIL and encrypted mempools in Hegota. What Glamsterdam does is build the protocol-level scaffolding required for those subsequent fixes to work.