Ethereum's Glamsterdam hard fork, targeting H1 2026 activation, is the network's most structurally significant upgrade since The Merge. Two headliner EIPs — EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists) — move block construction on-chain and enable paral...
"The resulting gas limit increase will be phased, reaching 100 million per block initially and 200 million once ePBS is fully operational." — Tomasz Stańczak, Co-Director, Ethereum Foundation
Ethereum's Glamsterdam hard fork, targeting H1 2026 activation, is the network's most structurally significant upgrade since The Merge. Two headliner EIPs — EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists) — move block construction on-chain and enable parallel transaction execution, respectively. The fork includes up to 22 EIPs and targets a phased gas limit increase from the current 60 million to 200 million per block, a 71% reduction in base ETH transfer costs, and a throughput path toward 10,000 TPS.
As of late April 2026, core developers have launched the first generalized devnet merging all Glamsterdam components into a single test environment. Public testnet activation on Holesky and Sepolia is expected in the coming weeks, with mainnet deployment in June possible but Q3 increasingly likely. The upgrade addresses a structural dependency: the top three block builders currently control over 80% of all PBS blocks, and more than 50% of high-value transactions route through private channels to avoid MEV extraction.
Ethereum's block production pipeline relies on an off-protocol system called MEV-Boost, operated by third-party relays. Validators outsource transaction packaging to external builders who bid for block inclusion rights. As of April 2026, Titan Builder holds approximately 52% of block production and BuilderNet (Flashbots' decentralized successor) holds roughly 25%. The former dominant builder, Beaverbuild, has declined to under 2% after migrating operations to BuilderNet in late 2024.
This relay dependency creates three problems. First, the network's liveness depends on infrastructure it does not control. Second, private transaction routing — now exceeding 50% of high-value transactions according to academic studies — concentrates MEV extraction among a small number of actors. Third, block construction remains sequential: each transaction executes in order, limiting throughput regardless of hardware improvements.
Ethereum currently operates at roughly 60 million gas per block. The network hosts over 1.1 million active validators securing approximately 35.8 million staked ETH (29.6% of circulating supply), valued at approximately $80 billion at current prices near $2,250. Annual MEV extraction is estimated at over $3 billion, with searchers extracting approximately $24 million in profit over a 30-day period between December 2025 and January 2026 alone.
EIP-7732 codifies the proposer-builder separation directly into Ethereum's consensus layer. Under the current system, a validator selected as block proposer outsources payload construction to a builder via MEV-Boost relays. The proposer commits to the highest-bidding builder's block header without seeing its contents, trusting the relay to verify validity.
Under ePBS, the protocol itself mediates this handoff. The proposer commits to a block header on-chain, and a separate builder constructs the execution payload without external relay infrastructure. The propagation window expands from 2 seconds to approximately 9 seconds, giving builders more time to construct optimal blocks while reducing the advantage of geographic proximity to relay servers.
The change eliminates what researchers call the "free option problem" — the window during which a builder can renege on a committed block. Academic modeling estimates this affects roughly 0.82% of blocks on average, rising to approximately 6% during periods of high volatility.
Critically, ePBS does not solve builder centralization. As Vitalik Buterin wrote in a February 2026 post: "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?" The upgrade isolates the problem — separating builder power from validator power — without resolving it. Buterin has pointed to "big FOCIL" (forced inclusion lists) and encrypted mempools as subsequent countermeasures.
EIP-7928 introduces Block-Level Access Lists (BALs), a mechanism that requires each block to carry a map of all accounts and storage slots it will access, along with post-execution state diffs. This converts Ethereum's execution model from a sequential process into a dependency graph.
The practical effect: nodes can identify which transactions access disjoint storage slots and execute them simultaneously. Historical data shows 60-80% of transactions access non-overlapping state, enabling significant parallelization. The remaining 20-40% with overlapping dependencies can still be partially parallelized using post-transaction state diffs included in the BAL.
BALs enable four specific optimizations: parallel disk reads across transactions, parallel transaction validation, parallel state root computation, and executionless state updates for nodes that receive the BAL before execution completes. Average BAL size is approximately 70 KiB per block based on historical modeling.
A companion proposal, EIP-8159, defines the peer-to-peer networking protocol (eth/71) for sharing access lists across nodes. Without this, BALs would increase bandwidth requirements without a standardized distribution mechanism.
Beyond the two headliners, Glamsterdam includes several EIPs that adjust Ethereum's fee structure and validator operations:
Gas repricing (EIP-8037, EIP-8038, EIP-2780): State creation gas costs increase to reflect actual storage burden, while intrinsic transaction gas decreases by approximately 71% for basic ETH transfers. The net effect: simple transfers become cheaper, but operations that expand Ethereum's state database become more expensive. The goal is to constrain annual state growth to approximately 100 GiB, down from roughly 200 GiB under current pricing.
Validator operations (EIP-8045, EIP-8080): Slashed validators are excluded from block proposals, and the exit queue is restructured to use the consolidation queue, increasing exit throughput by up to 2.5x during high-demand periods. This is relevant given the 1.1 million active validators and the increasing validator balance ceiling raised to 2,048 ETH under Pectra.
Developer tooling (EIP-7997, EIP-7708, EIP-7975): A deterministic factory predeploy enables identical contract addresses across EVM chains. ETH transfers and burns now emit standard logs. Receipt pagination prevents sync failures during peak activity.
The gas limit trajectory represents Ethereum's most aggressive capacity expansion since genesis. The current 60 million limit — itself raised from 36 million in early 2025 — will increase to 100 million in the first phase. Once ePBS is fully operational and the extended propagation window stabilizes, the target rises to 200 million.
At 200 million gas per block, Ethereum's theoretical throughput approaches 10,000 TPS — roughly ten times its current effective rate. Stańczak described this as making mainnet "a solid rock and nimble network" that institutions can trust.
The throughput target intersects with Ethereum's L2 strategy. The Fusaka upgrade, activated in December 2025, introduced PeerDAS for data availability sampling. Glamsterdam's BALs complement this by making L1 execution itself more efficient. The combined effect: L1 handles more transaction volume directly while maintaining cheap data availability for L2s.
Long-term, the Ethereum Foundation has outlined ZK-proof validation as a subsequent scaling vector. The Hegota upgrade, planned for late 2026, is expected to advance this work. The 10,000 TPS figure represents a near-term ceiling, not a final state.
Development has progressed through several phases:
The critical next steps: stabilization of the generalized devnet, client releases from execution and consensus teams, dual security audits, and public testnet activation on Holesky and Sepolia. June mainnet activation remains possible but increasingly tight. Q3 2026 is the more probable window based on the current pace.
This would make Glamsterdam the third Ethereum hard fork within one year, following Pectra (May 2025) and Fusaka (December 2025).
The economic effects distribute across several stakeholder groups:
Validators and stakers retain block proposal rights but lose the option to independently construct profitable blocks — a function already outsourced by 90%+ of validators via MEV-Boost. The structural change is minimal for most operators. The expanded 2,048 ETH max effective balance from Pectra means fewer but larger validators can participate, and Glamsterdam's exit queue improvements accommodate this consolidation.
Block builders face the most direct impact. ePBS removes relay operators as intermediaries but preserves the builder auction model. Builders who currently benefit from private orderflow arrangements and geographic proximity to relays will see some advantages diminish. Titan's 52% market share may face pressure as protocol-level auctions reduce barriers to entry.
Application developers and users benefit from the 71% reduction in basic transfer costs and improved throughput. For DeFi protocols, parallel execution means reduced contention for state access during high-demand periods — a persistent source of gas spikes during market volatility.
L2 operators see indirect benefits from improved L1 finality guarantees and continued cheap data availability, though Glamsterdam's primary scaling impact is on L1 itself.
MEV dynamics shift but do not disappear. The approximately $3 billion annual MEV economy moves from relay-mediated extraction to protocol-mediated extraction. Whether total MEV increases or decreases depends on builder competition under ePBS — a question the data cannot yet answer.
Glamsterdam is an infrastructure upgrade, not a feature release. It restructures how blocks are built, how transactions are executed, and how gas is priced — the three foundational layers of Ethereum's economic model. The value proposition is not speed or cost reduction in isolation, but the removal of off-protocol dependencies that currently mediate the majority of block production.
The upgrade arrives as Ethereum's staking economy exceeds $80 billion in secured value, its L2 ecosystem processes the majority of user transactions, and institutional participants increasingly deploy capital on-chain. The question is not whether these structural changes are necessary — relay dependency and sequential execution are widely acknowledged bottlenecks — but whether the compressed development timeline produces a stable deployment.
Three hard forks in 13 months is unprecedented for Ethereum. The generalized devnet's performance over the coming weeks will determine whether June remains viable or Q3 becomes the realistic target.