Ethereum's Glamsterdam hard fork, targeted for mid-2026, represents the network's most substantial architectural overhaul since The Merge in September 2022. Two headliner proposals — EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists) — aim to restructure how ...
"In 2026, no longer. Every compromise of values that Ethereum has made up to this point — every moment where you might have been thinking, is it really worth diluting ourselves so much in the name of mainstream adoption — we are making that compromise no longer." — Vitalik Buterin, Ethereum Co-Founder
Ethereum's Glamsterdam hard fork, targeted for mid-2026, represents the network's most substantial architectural overhaul since The Merge in September 2022. Two headliner proposals — EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists) — aim to restructure how blocks are built, ordered, and executed at the protocol level.
The upgrade targets a gas limit increase from 60 million to 200 million per block, a throughput trajectory toward 10,000 TPS on the base layer (up from the current ~15–30 TPS), and an estimated 78% reduction in gas fees for complex smart contract interactions. As of late April 2026, the first generalized devnet merging all Glamsterdam components is live, though developers have signaled that a Q3 slip from the aspirational June target remains possible.
The economic stakes are substantial. Today, 80–90% of Ethereum block production depends on off-chain builders operating through external relay infrastructure. Two builders — Titan and BuilderNet — construct approximately 80% of all blocks. Cumulative MEV extraction on Ethereum has surpassed $1.3 billion. Glamsterdam's ePBS proposal moves this entire pipeline on-chain, eliminating the need for trusted intermediaries and standardizing MEV handoff rules at the consensus layer.
Ethereum's block production today relies on an off-chain market known as MEV-Boost, originally introduced by Flashbots after The Merge. In this system, specialized builders assemble blocks to capture maximal extractable value, then submit bids to proposers (validators) through relay intermediaries.
The concentration is acute. According to data from Rated Network, Titan Builder controls approximately 52–57% of blocks, while Flashbots' BuilderNet handles roughly 23–25%. The top three builders collectively produce over 80% of all PBS blocks, according to Bitfinex Research. This creates a single-point-of-failure risk: if two or three entities go offline or collude, Ethereum's liveness and censorship resistance degrade.
Relay infrastructure adds a separate trust layer. Ultra Sound relay handles 32.3% of MEV-Boost payloads, Titan Relay 24.75%, and the two bloXroute relays combined approximately 26%, according to relayscan.io data from late 2025. Validators must trust that relays faithfully transmit blocks — a trust assumption not enforced by the protocol.
MEV extraction continues to grow. Cumulative MEV profits on Ethereum surpassed $1.3 billion as of early 2026, with roughly $40–60 million captured monthly, concentrated in arbitrage and sandwich strategies, according to data compiled by ESMA. Flashbots Protect had saved users over 4,600 ETH in MEV and 2,200 ETH in gas fees by March 2026, but this represents a fraction of total extraction.
EIP-7732 moves the proposer-builder handoff directly into the Ethereum consensus protocol. The design introduces a commit-reveal flow: builders cryptographically seal blocks and commit to bids; validators select the highest bid without viewing transaction contents; a new Payload Timeliness Committee (PTC) validates that the builder reveals the payload on time.
Three structural changes result from this:
Relay elimination. The trusted relay layer becomes unnecessary. Block commitments and reveals are enforced at the consensus level, with penalties for non-compliance. If a builder fails to reveal its payload on time, it forfeits its full bid value. The proposer retains bid payment regardless of builder behavior.
Expanded propagation window. Under the current system, validators have approximately 2 seconds to propagate data. EIP-7732 extends this window to roughly 9 seconds, enabling the network to handle larger blocks and more data blobs for Layer 2 networks.
MEV fairness. By standardizing the handoff rules on-chain, the proposal eliminates the information asymmetry that allows relays to front-run or censor transactions. Some researchers estimate MEV extraction could be reduced by up to 70% under ePBS, though this figure is contested.
A critical caveat: academic modeling indicates ePBS may "significantly amplify profit and content centralisation" among builders who maintain private order flow advantages, according to research cited by Bitfinex. The free option problem — where builders withhold blocks after bidding in approximately 0.82% of cases (rising to 6% during volatility) — remains an unresolved design tension.
The second headliner addresses execution-layer throughput. Today, Ethereum processes transactions sequentially within each block. EIP-7928 introduces Block-Level Access Lists (BALs) — a record of every account, storage slot, balance, nonce, and code change accessed during block execution.
BALs enable four capabilities:
A new field, block_access_list_root, is added to the block header. Nodes must retain BALs for a minimum of 3,533 epochs. A new protocol version, eth/71, handles peer-to-peer BAL distribution. According to research published on ethresear.ch, worst-case parallel execution modeling under EIP-7928 shows substantial throughput gains even in adversarial scenarios.
The throughput architecture combines BALs with a phased gas limit increase. According to former Ethereum Foundation co-executive director Tomasz Stanczak, speaking at the Bankless Summit, the gas limit will increase to 100 million per block initially, then double to 200 million once ePBS is fully operational. The current gas limit stands at 60 million.
This 3.3x gas limit increase, combined with BAL-enabled parallel execution, targets approximately 10,000 TPS on the base layer — up from the current effective rate of 15–30 TPS. For context, Ethereum's current ecosystem throughput averages approximately 325 TPS when including Layer 2 scaling networks, according to Datawallet.
Gas fees, already at historic lows (0.052 Gwei average in April 2026, down from 1.67 Gwei a year earlier, per CoinLaw), are projected to fall a further 78% for complex smart contract interactions. This reduction comes partly from gas repricing EIPs under consideration.
Beyond the two headliners, four additional EIPs are under consideration for Glamsterdam inclusion:
Several proposals were explicitly declined: FOCIL inclusion lists (deferred to future forks), reduced slot times, multidimensional gas metering, and post-quantum signature verification. The scope freeze locked the technical ingredients in late Q1 2026, per developer documentation.
Development has proceeded through five devnet iterations. Devnet-4 testing is complete; Devnet-5 was underway as of April 2026. Critically, the first generalized devnet — merging ePBS and BAL components into a single test environment — launched in the final week of April 2026. Prior to this, testing was split across separate networks.
Ethereum Foundation Checkpoint #9, published April 10, 2026, described the pace as "slow but steady," identifying ePBS as the primary bottleneck due to the complexity of coordinating two parties within consensus. The transition from Devnet-5 to public testnet represents the next major milestone, with dual audit rounds planned before any mainnet deployment.
The June 2026 target remains aspirational. Multiple sources, including core developer commentary, suggest Q3 2026 is more realistic. Ethereum has a documented history of delaying major upgrades; The Merge itself was postponed multiple times over several years.
This fits a broader upgrade cadence: Pectra activated in May 2025, Fusaka in December 2025, followed by two blob parameter optimization (BPO) forks in December 2025 and January 2026 that tripled blob capacity.
Builder centralization persistence. ePBS eliminates relay trust assumptions but does not directly address builder-level concentration. Builders with superior private order flow may entrench their dominance even further under the new regime.
Validator hardware requirements. With a 200 million gas limit, the computational burden on validators increases substantially. The Ethereum Foundation's roadmap anticipates that a large number of validators will eventually shift from re-executing transactions to verifying zero-knowledge proofs, but this transition is not part of Glamsterdam.
Free option risk. The 0.82% block withholding rate (6% during volatility) under ePBS remains an open design question. Penalty mechanisms exist, but their calibration for high-volatility conditions is unresolved.
L2 dependency. Layer 2 networks account for approximately 95% of Ethereum's total transaction throughput. While Glamsterdam improves L1 capacity, the structural dependency on L2 economics — and whether L1 fee reduction further erodes L1 revenue — creates tension with Ethereum's long-term value accrual model.
Staking concentration. With over 1.1 million active validators and 35.9 million ETH staked (28.9% of total supply, reaching 31.1% in March 2026 per IndexBox), the validator set is large but increasingly institutionalized. Larger staking pools stand to benefit disproportionately from the operational changes in block building.
For the 1.1 million validators securing approximately $112 billion in staked ETH (at January 2026 prices), Glamsterdam restructures the revenue model. Proposers gain a more predictable income stream from builder bids, enforced by protocol-level guarantees rather than relay trust. The average staking yield of 3.3% annually may shift as MEV redistribution mechanics change.
For DeFi users transacting on L1, a 78% gas fee reduction on complex operations — combined with already-low fees of approximately $0.01 per transaction — reduces execution costs toward negligibility. This narrows the economic rationale for L2 migration on cost grounds alone, potentially rerouting some transaction volume back to L1.
For Layer 2 operators, the expanded data propagation window under ePBS and continued blob scaling provide cheaper data availability. Combined with the Fusaka-era PeerDAS upgrade that cut per-node bandwidth by 90%, rollup operating costs continue to compress.
Glamsterdam is not a performance patch. It is a structural rewrite of how Ethereum's base layer builds, orders, and executes blocks. The upgrade directly addresses the tension Buterin identified: years of pragmatic compromise — outsourcing block building to off-chain relays, accepting builder concentration, tolerating sequential execution — in exchange for faster adoption.
The economic question is whether the protocol can reclaim value currently extracted by intermediaries without entrenching the same intermediaries under a new regime. Academic concerns about ePBS amplifying builder centralization suggest the answer is uncertain. The data from live devnets will determine whether Ethereum's largest architectural bet since The Merge delivers on its stated objectives.