Ethereum's next hard fork, Glamsterdam, targets activation in H1 2026 with two protocol-level changes that would restructure how blocks are built and processed on the network. EIP-7732 (Enshrined Proposer-Builder Separation) moves the block-building market from third-party relays into the consens...
Ethereum's next hard fork, Glamsterdam, targets activation in H1 2026 with two protocol-level changes that would restructure how blocks are built and processed on the network. EIP-7732 (Enshrined Proposer-Builder Separation) moves the block-building market from third-party relays into the consensus layer. EIP-7928 (Block-Level Access Lists) enables parallel transaction execution by making storage-slot dependencies explicit at the block level. Together, the proposals aim to raise the gas limit from 60 million to 200 million per block, target throughput of 10,000 transactions per second—roughly 10x current effective capacity—and reduce gas fees by approximately 78%.
The upgrade addresses a structural problem: three block builders currently control more than 80% of all PBS blocks on Ethereum. MEV-Boost relay dominance has concentrated economic power off-chain, outside protocol governance. Glamsterdam attempts to bring this market on-chain while simultaneously scaling Layer 1 execution. However, academic modeling suggests ePBS may amplify builder-side profit concentration even as it reduces validator-side centralization. The Ethereum Foundation's DevOps team has completed Devnet-4 testing and is transitioning to Devnet-5, but neither ePBS nor BALs have been proven at mainnet scale.
Ethereum's proposer-builder separation system operates primarily through off-chain relays. According to Cointelegraph, two builders—Beaverbuild and Titan Builder—were responsible for 88.7% of all blocks produced on mainnet in a recent two-week sample period. The top three builders produce more than 90% of blocks. Approximately 60% of block value derives from private order flow, according to research from Decentralized Thoughts, creating an implicit access barrier that reinforces incumbency.
The annual value extracted through MEV on Ethereum exceeded $3 billion as of 2025, according to data compiled by CalmOps. Flashbots, which historically relayed over 82% of MEV-Boost blocks, migrated all builders to its BuilderNet architecture in December 2024—a joint operation with Beaverbuild and Nethermind. This shift toward decentralized relay infrastructure has not, however, materially altered the concentration of block production among top builders.
Over 50% of high-value transactions as of mid-2025 were routed through private channels to avoid MEV extraction, according to Bitfinex. The current system functions, but it operates largely outside protocol governance—a structural dependency the Ethereum Foundation has identified as a long-term risk.
EIP-7732 moves the proposer-builder separation mechanism from off-chain relays into Ethereum's consensus layer. Under the current system, validators running MEV-Boost software outsource block construction to specialized builders who compete in off-chain auctions. Under ePBS, builders cryptographically seal block contents and submit bids directly on-chain. Proposers select the highest-paying bid without viewing transaction contents before commitment locks.
The design eliminates the network's current 80–90% reliance on third-party relay infrastructure, according to multiple sources including CCN and BingX. It creates a permissionless, protocol-native marketplace for block construction.
However, the mechanism introduces the "free option problem." A builder can withhold block payload after committing to a bid if late-arriving MEV makes abandoning it more profitable. Academic modeling estimates this affects approximately 0.82% of blocks on average, rising to roughly 6% during periods of high volatility, according to analysis cited by Bitfinex.
A January 2026 academic paper modeling ePBS in the presence of MEV found that while the mechanism reduces validator-side concentration, it "significantly amplifies profit and content centralisation" among builders, because access to private order flow confers a structural bidding advantage that compounds over time. The implication: ePBS may solve one centralization vector while intensifying another.
To address this, Vitalik Buterin proposed supplementary measures including FOCIL (Fork-Choice Obligatory Commitment to Inclusion Lists), which would select 16 random attesters per slot to mandate transaction inclusion, and encrypted mempools to obscure transaction contents from builders until after block commitment.
EIP-7928 introduces Block-Level Access Lists (BALs), which make storage-slot access patterns explicit at the block level. This enables the Ethereum Virtual Machine to identify which transactions touch non-overlapping state and execute them simultaneously rather than sequentially.
Current Ethereum execution is serial: each transaction processes one after another. BALs provide the metadata layer necessary for parallel execution without sacrificing deterministic state transitions. The gas limit increase from 60 million to 200 million provides the raw block capacity; BALs provide the intelligence to utilize that capacity efficiently.
The combination targets effective throughput of approximately 10,000 TPS, up from the current effective rate of roughly 1,000 TPS on the base layer. Gas repricing via EIP-7904, also included in the Glamsterdam scope, contributes to the projected 78% fee reduction across both simple transfers and complex smart contract interactions.
At current gas prices of approximately 0.16 gwei (per Etherscan data for April 2026), a simple ETH transfer costs roughly $0.006 at an ETH price near $1,920. If Glamsterdam delivers the projected 78% fee reduction, that figure drops further—though the practical significance at already sub-cent levels may be limited for simple transfers. The throughput gains matter more for DeFi and smart contract interactions, where gas costs scale with execution complexity.
Former Ethereum Foundation co-executive director Tomasz Stańczak indicated a phased gas limit increase: an initial jump to 100 million per block, followed by a post-ePBS expansion to 200 million. Vitalik Buterin outlined the full scope in eight EIPs published in late February 2026.
The 200-million gas target represents a 3.3x increase from the current 60-million limit. Combined with parallel execution via BALs, the multiplicative effect on throughput is projected at approximately 10x. However, this assumes validator nodes can handle the increased computational and bandwidth requirements within the network's decentralization constraints.
Over 25 non-headliner EIPs are under consideration for inclusion in the Glamsterdam scope. Base's engineering team has publicly warned that adding FOCIL alongside ePBS could delay the upgrade beyond 2026, raising questions about scope management.
Glamsterdam expands the number of data blobs available per block—potentially up to 72 or more—giving optimistic and ZK rollups additional capacity to post compressed transaction data to Ethereum. More blob capacity reduces bidding wars for data availability space, which should lower Layer 2 fees without requiring mainnet to process every transaction.
This matters economically. Base currently generates the highest L2 revenue at approximately $185,291 per day, with priority fees forming 86.1% of its income. Arbitrum One leads in TVL at $15.2 billion and rolled out dynamic pricing in January 2026. According to 21Shares, most smaller L2s may not survive 2026 as Base, Arbitrum, and Optimism tighten their grip on the market.
The Fusaka upgrade, shipped December 3, 2025, already initiated blob scaling via PeerDAS with a maximum target of 48 blobs per block. Glamsterdam extends this further. Optimism has modeled an upper-end scenario where 48+ blob targets per block, paired with rollup throughput improvements, move effective capacity from approximately 220 to roughly 3,500 user operations per second.
If executed as designed, users interacting with rollups could see fees at fractions of a cent while maintaining Ethereum's security guarantees. The economic question is whether expanded blob capacity drives net new demand or simply compresses margins for existing L2 operators.
Glamsterdam introduces material changes to validator duties. Under ePBS, validators gain membership in a new Payload Timeliness Committee (PTC), responsible for verifying that builders reveal block contents on time. This adds monitoring requirements and timing sensitivities that do not exist in the current protocol.
Approximately 37 million ETH is locked in staking—30.6% of circulating supply—with over 3 million validators in the entry queue. The Ethereum Foundation itself staked 45,034 ETH on April 3, 2026, pushing its total to roughly 69,500 ETH.
The broader validator risk profile extends beyond Glamsterdam. Prysm, the dominant consensus client, still controls over one-third of staked ETH, creating a single point of failure. The eventual transition from block re-execution to ZK execution proof verification—planned for the Hegota upgrade in late 2026—will further alter validator economics and operational requirements.
Delay risk. The tentative June 2026 target could slip to Q3 or Q4. Over 25 non-headliner EIPs under consideration for inclusion expand the scope and testing surface. The Base team's warning about FOCIL inclusion is one public signal of scope pressure.
Builder centralization transfer. ePBS addresses relay centralization but may intensify builder centralization. Builders with access to private order flow retain a structural advantage that protocol-level changes alone do not eliminate.
Untested at scale. Neither ePBS nor BALs have been proven at mainnet scale. Devnet-4 is complete and Devnet-5 is underway, but the interplay between these two features under real network conditions remains unvalidated. Public testnet activations on Holesky and Sepolia are expected in the months preceding mainnet launch.
Consensus-layer complexity. The upgrade substantially increases consensus-layer complexity, broadening the attack surface for bugs and edge cases. The combination of parallel execution, on-chain block auctions, and expanded blob space represents three simultaneous structural changes.
L1 vs. L2 demand uncertainty. Whether network demand materializes as blob usage (favoring L2s) versus L1 execution bidding (favoring direct mainnet activity) remains an open question that will shape fee dynamics and economic value distribution post-upgrade.
Glamsterdam represents the most structurally ambitious Ethereum upgrade since The Merge. It attempts to solve three problems simultaneously: block builder centralization, base-layer throughput limitations, and gas fee levels. The economic stakes are substantial—over $3 billion in annual MEV flows through a system currently governed by three dominant builders operating outside protocol rules.
The technical design is coherent: ePBS brings the builder market on-chain while BALs unlock parallel execution. But coherent design and successful mainnet deployment are different things. The upgrade has not been tested at scale, the scope continues to expand, and the centralization problem it aims to solve may partially reproduce itself in a new form. The data suggests Glamsterdam is necessary infrastructure work. Whether it ships on time and delivers its projected improvements without introducing new systemic risks is the open question worth tracking through H1 2026.