Ethereum's next hard fork, Glamsterdam, is targeting a May–June 2026 mainnet activation. The upgrade centers on two headliner Ethereum Improvement Proposals: EIP-7732 (Enshrined Proposer-Builder Separation, or ePBS) and EIP-7928 (Block-Level Access Lists). Together, they represent the most signif...
"In Glamsterdam, Ethereum is getting ePBS, which lets proposers outsource to a free permissionless market of block builders. This ensures that block builder centralization does not creep into staking centralization, but it leaves the question: what do we do about block builder centralization?" — Vitalik Buterin, Ethereum Co-Founder (March 2, 2026)
Ethereum's next hard fork, Glamsterdam, is targeting a May–June 2026 mainnet activation. The upgrade centers on two headliner Ethereum Improvement Proposals: EIP-7732 (Enshrined Proposer-Builder Separation, or ePBS) and EIP-7928 (Block-Level Access Lists). Together, they represent the most significant protocol-level restructuring of Ethereum's block production pipeline since The Merge in September 2022.
The stakes are material. Approximately 90% of all Ethereum blocks are currently built by four entities, with Beaverbuild alone commanding nearly 50% market share. MEV extractors captured an estimated $24 million in a single 30-day period between December 2025 and January 2026. Glamsterdam moves the proposer-builder coordination layer — currently handled by off-chain relays like MEV-Boost — directly into the protocol, while enabling parallel transaction execution through enforced state-access declarations. Researchers estimate ePBS could reduce MEV extraction by up to 70%.
The upgrade carries execution risk. Devnet-4 testing is complete; the Foundation's DevOps team is transitioning to Devnet-5. Over 50 EIPs were initially proposed; developers narrowed scope to roughly 17 non-headliner features alongside the two headliners. FOCIL (Fork-Choice Inclusion Lists), a censorship-resistance mechanism, was deliberately excluded from Glamsterdam after the Base engineering team warned it could delay the fork beyond 2026. It has been deferred to the subsequent Hegotá upgrade.
Since the introduction of MEV-Boost after The Merge, Ethereum has operated a de facto proposer-builder separation system — but one that relies on trusted, off-chain relay infrastructure. The result has been acute concentration. According to Etherscan data, just two builders — Beaverbuild and Titan Builder — construct approximately 88% of all Ethereum blocks. Four builders account for roughly 90%.
This concentration creates measurable problems beyond mere aesthetics. Both Beaverbuild and rsync-builder have been documented censoring OFAC-sanctioned transactions, meaning only about 40% of Ethereum blocks are fully censorship-resistant. Academic research published in mid-2025 found that nearly 46% of blocks are produced by actors explicitly enforcing OFAC-based or other censorship policies.
MEV extraction remains substantial. Between December 8, 2025, and January 6, 2026, MEV extractors generated approximately $24 million in 30 days on Ethereum mainnet. Sandwich attacks constituted $289.76 million, or 51.56%, of total MEV transaction volume in a measured period of $561.92 million. Validators capture the largest share: empirical data from early 2023 showed validators receiving 72.1% of extracted MEV ($30.3 million out of $48.3 million in a two-month sample), with searchers taking 17.4% and builders 10.5%.
Beaverbuild announced in May 2025 that it would retire its centralized builder and transition users to BuilderNet, a decentralized block-building platform developed with Flashbots. The move acknowledged what protocol researchers had long argued: relay-dependent PBS creates single points of failure and trust assumptions that contradict Ethereum's design goals.
EIP-7732 moves the entire proposer-builder coordination mechanism on-chain. Under the current MEV-Boost system, proposers receive block headers from builders via external relays, select the highest-bidding header, and commit to it. The relay acts as a trusted intermediary. Under ePBS, this process becomes part of the consensus protocol itself.
The mechanism works as follows: the proposer commits to a block header on-chain. A separate builder constructs the execution payload. No external relay is required. The block-building logic becomes enshrined in Ethereum's consensus layer. If a builder fails to deliver a valid payload matching the committed header, the protocol handles the failure natively rather than relying on relay-level fallback mechanisms.
The economic implications are direct. By removing relays as intermediaries, ePBS eliminates a rent-extraction layer and reduces trust assumptions. Researchers estimate the change could reduce MEV extraction by up to 70%, translating to fairer execution for traders, borrowers, and liquidity providers on Ethereum's base layer. The 261% increase in proposer revenue observed after the introduction of MEV-Boost PBS suggests that protocol-level formalization could further redistribute value toward validators.
Critically, ePBS does not solve builder centralization — it formalizes the market. As Buterin noted in his March 2, 2026, blog post, enshrining PBS prevents builder centralization from infecting the staking layer, but the question of who builds blocks remains open.
EIP-7928 introduces Block-Level Access Lists (BALs), which record all accounts and storage locations accessed during block execution, along with their post-execution values. A new field, block_access_list_hash, is added to the block header containing the Keccak-256 hash of the RLP-encoded access list.
The practical consequence: parallel transaction execution. Currently, Ethereum processes transactions sequentially because the EVM cannot know in advance which state slots a transaction will read or write. EIP-7928 solves this by requiring that state access be declared upfront at the block level. Only addresses and storage slots actually touched during execution are recorded. The access list is constrained by the block gas limit rather than a fixed maximum number of items.
This enables four specific optimizations: parallel disk reads, parallel transaction validation, parallel state root computation, and executionless state updates. While EIP-2930 introduced optional transaction-level access lists in 2021, they were never enforced. EIP-7928 makes them mandatory at the block level.
The performance targets are ambitious. According to multiple developer sources, Glamsterdam aims for throughput of 10,000 TPS and a 78.6% reduction in gas fees across both simple transfers and complex smart contract interactions. Current mainnet gas prices already sit at historic lows — Etherscan shows an average of 0.058 gwei as of late March 2026, with ERC-20 transfers costing $0.01–$0.02 and Uniswap swaps approximately $0.14. Further reductions would push Ethereum L1 costs closer to L2 territory.
Fork-Choice Inclusion Lists (FOCIL) was one of the most debated proposals during Glamsterdam scoping. The mechanism would require a randomly selected committee of 16 validators to create a list of transactions that must be included in the next block. If a builder attempts to exclude these transactions, the block would be rejected by the network.
FOCIL directly addresses the censorship problem: if 46% of blocks enforce OFAC-based filtering, a protocol-level mandate for transaction inclusion would override builder-level censorship decisions. The mechanism received strong support from core developers and the broader Ethereum community.
However, the Base engineering team publicly warned that bundling FOCIL with ePBS could delay Glamsterdam beyond 2026. The concern was scope creep: the interplay between ePBS, BALs, and FOCIL introduced untested complexity at mainnet scale. Core developers opted to exclude FOCIL from Glamsterdam and designate it as the first confirmed headliner for Hegotá, the subsequent hard fork expected in H2 2026.
Buterin has since proposed an extended version — "Big FOCIL" — that would expand the inclusion list's capacity to handle nearly all transactions in a block, effectively reducing the builder's role to state computation and residual MEV optimization. The proposal also includes encrypted mempools to prevent front-running, a mechanism targeting what researchers term "toxic MEV."
The Ethereum Foundation's DevOps team has completed testing of three proposed EIPs on Devnet-4. The transition to Devnet-5 is underway. According to developer reports from late February 2026, the Lodestar and Lighthouse consensus clients have achieved interoperability, with the goal of supporting 3–4 consensus-layer clients. Perf-Devnet-2 has resumed operations, with Erigon conducting snapshot analysis post-synchronization. Gas-limit testing focused on Glamsterdam-specific repricing benchmarks is ongoing.
The tentative mainnet target is May–June 2026. Developers have repeatedly emphasized that this timeline depends on successful testnet validation across all environments. Given the upgrade's scope — larger than any fork since The Merge — slippage to Q3 or Q4 2026 remains plausible. Public testnets are expected before mainnet activation.
Of the initial 50+ non-headliner EIPs proposed, developers narrowed the scope to approximately 17 features. More than 30 proposals were explicitly excluded. The two headliners (EIP-7732 and EIP-7928) carry the bulk of the upgrade's complexity.
Glamsterdam arrives as Ethereum navigates a period of structural tension. ETH trades near $1,990 as of March 29, 2026 — below the $2,000 psychological level that has served as contested support. The network is secured by approximately 1.1 million active validators staking 35.86 million ETH (28.91% of total supply), representing roughly $112 billion in economic security as of early January 2026.
L2 activity continues to accelerate relative to L1. Daily transactions on Layer 2 networks exceeded 1.9 million per day during 2025, eclipsing mainnet. Projections from multiple analyst firms estimate L2 TVL could surpass L1 DeFi TVL by Q3 2026 — $150 billion versus $130 billion on mainnet. L2 sequencer revenue runs approximately $161 million annually.
The Ethereum Foundation published a strategic framework on March 23, 2026, explicitly defining L1's role: to "serve as a truly permissionless and maximally resilient global hub for settlement, shared state, liquidity, and DeFi." L2s, by contrast, should provide "differentiated features, services, customizations, go-to-market strategies, and zones of control." Glamsterdam's performance improvements directly serve this vision by making L1 more competitive as a settlement layer.
Gas fees at 0.058 gwei represent a paradox. Usage is near all-time highs, but fee revenue to the protocol has collapsed. The Dencun (2024) and Fusaka (2025) upgrades slashed data costs to the point where the network's economic model — in which validators are compensated partly through fee burns — faces questions about long-term sustainability. Whether Glamsterdam's throughput gains drive enough additional volume to offset per-transaction fee declines remains to be seen.
Scope risk. Glamsterdam's combination of ePBS and BALs introduces protocol-level complexity that has not been tested at mainnet scale. The decision to exclude FOCIL reduced scope but did not eliminate the fundamental challenge of shipping two large, interacting EIPs simultaneously.
Builder behavior post-ePBS. Enshrining PBS eliminates relay trust assumptions but does not guarantee builder decentralization. If the same entities that dominate MEV-Boost (Beaverbuild, Titan) continue to dominate the on-chain builder market, censorship patterns may persist. FOCIL's deferral to Hegotá leaves a gap.
Validator operational burden. The shift from 32 ETH to 2,048 ETH maximum effective balance (introduced in Pectra) combined with ePBS's new consensus responsibilities may increase hardware and bandwidth requirements for validators, potentially favoring institutional operators.
Timeline slippage. The June target is contingent on Devnet-5, public testnet, and multi-client interoperability milestones. Any issues in consensus-execution layer interaction could push mainnet activation to Q3–Q4 2026.
Glamsterdam represents Ethereum's most consequential protocol change since The Merge. By enshrining proposer-builder separation, the network eliminates its dependency on trusted relay infrastructure — a structural vulnerability that has enabled both builder centralization and transaction censorship. Block-Level Access Lists, meanwhile, unlock parallel execution capabilities that have been discussed in Ethereum research circles for years.
The trade-off is clear: Glamsterdam addresses the architecture of block production but deliberately defers the question of censorship resistance to a later upgrade. For an estimated 46% of blocks currently subject to OFAC filtering, relief will not arrive until FOCIL ships in Hegotá. Whether the builder market decentralizes under ePBS or merely formalizes existing power structures remains an empirical question that only mainnet data can answer.
The next three months of testnet validation will determine whether Ethereum meets its most ambitious technical deadline since the transition to proof of stake.