Ethereum's Glamsterdam hard fork, targeted for June 2026, enters Devnet-5 testing as the network's most structurally significant upgrade since The Merge. The centerpiece — EIP-7732, or enshrined Proposer-Builder Separation (ePBS) — moves block construction into the protocol itself, replacing the ...
"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
Ethereum's Glamsterdam hard fork, targeted for June 2026, enters Devnet-5 testing as the network's most structurally significant upgrade since The Merge. The centerpiece — EIP-7732, or enshrined Proposer-Builder Separation (ePBS) — moves block construction into the protocol itself, replacing the off-chain relay infrastructure that currently handles 80–90% of block production through informal trust agreements.
The upgrade arrives at a moment of acute concentration in Ethereum's block-building market. Three builders — Beaverbuild (35.7%), Titan Builder (23.2%), and Rsync (12.9%) — account for roughly 72% of all MEV-Boost blocks. The top ten builders produce 97.7% of blocks and capture 83.9% of the estimated $52 million in cumulative builder profit. Glamsterdam does not eliminate this concentration, but it erects protocol-level walls preventing it from leaking into validator centralization.
Alongside ePBS, EIP-7928 introduces Block-Level Access Lists and EIP-7904 proposes benchmarked gas repricing. Combined, the package targets a 10–30x improvement in execution-layer throughput in the near term, with the gas limit rising from 60 million to 200 million per block. ETH trades at approximately $2,119 as of April 6, 2026, with average gas prices at 0.09 gwei — levels already compressed by the 2024 Dencun upgrade.
Ethereum's current block production operates through MEV-Boost, a system Flashbots introduced in 2022 to separate the roles of block proposers (validators) and block builders. Validators outsource block construction to specialized builders who compete in off-chain auctions managed by relay operators. The system works but is not enshrined in the protocol — it runs on trust, off-chain coordination, and voluntary participation.
The resulting market structure is oligopolistic. Between October 2023 and March 2024, three builders produced 80% of all MEV-Boost blocks: Beaverbuild (34.9%), Rsync (23.0%), and Titan (22.7%), according to data compiled by researchers at Decentralized Thoughts and corroborated by Rated Network's builder explorer. By Q1 2026, Beaverbuild holds 35.7% market share with over 2.1 million blocks, followed by Titan at 23.2% (1.38 million blocks), Rsync at 12.9% (769,000 blocks), and Flashbots at 9.4% (563,000 blocks).
Barriers to entry are significant. Academic research published in July 2024 (arXiv:2407.13931) found that profitability correlates directly with access to private order flow from exclusive providers — integrated searchers and external partners with exclusivity deals. Private order flows contribute 54.6% of block value, meaning builders with more exclusive deal flow win more auctions, compounding their dominance. The cost to establish even a 1% market share rose to 1.4 ETH by March 2024.
Flashbots attempted to address this in late 2024 with BuilderNet, a collaborative block-building network co-operated with Beaverbuild and Nethermind. By December 2024, Flashbots ceased operating centralized builders entirely and migrated all activity to BuilderNet. The initiative targets exclusive orderflow neutralization and MEV distribution, but operates outside the protocol layer — making it a voluntary coordination mechanism rather than a consensus-enforced rule.
EIP-7732 restructures the validator lifecycle to separate consensus validation from execution validation, both logically and temporally. Under the current system, a validator proposing a block must either build it themselves or trust an off-chain relay to deliver a builder's payload. Under ePBS, the process changes:
Commitment phase. The ExecutionPayload field is removed from the BeaconBlockBody and replaced with a SignedExecutionPayloadHeader — a cryptographic commitment from a builder specifying the blockhash and a bid (value to be paid to the proposer). Builders are staked on the beacon chain, making their bids enforceable without relying on external trust.
Selection. The proposer selects the highest bid from the on-chain auction without seeing the transaction contents of the block. This is the sealed-bid mechanism: proposers optimize for revenue, builders compete on execution quality and MEV capture.
Revelation. After the commitment is locked, the builder reveals the full execution payload. A new validator duty — payload timeliness attestation — ensures the payload is delivered on time. If the builder fails to reveal, their stake is slashed and the proposer is still paid.
Settlement. The committed value is deducted from the builder's beacon chain balance and credited to the proposer, all within the consensus layer. No relay, no off-chain escrow, no trust assumption.
The result: validators need only process block headers and select from on-chain bids. They no longer need to run MEV software, negotiate with relays, or trust any particular builder. The relay layer — currently a single point of failure and censorship vector — becomes unnecessary for the proposer-builder handoff.
What ePBS does not do is equally important. It does not reduce builder concentration. The same builders with exclusive order flow will likely continue winning auctions. What it prevents is that concentration from migrating upstream: a dominant builder cannot leverage their position to also dominate validation, because the protocol enforces a clean separation. As Buterin stated in his March 7, 2026 post outlining Ethereum's roadmap priorities, ePBS is a containment measure, not a cure.
EIP-7928: Block-Level Access Lists (BALs). This proposal introduces a deterministic record of accounts and storage slots accessed during block execution. BALs enable parallel transaction processing by identifying which state elements a block touches, allowing client software to pre-fetch and parallelize reads. Archival nodes must store BALs for at least 3,533 epochs. The Ethereum Foundation DevOps team has been testing BALs on bals-devnet-2 as of Q1 2026.
EIP-7904: General Repricing. This EIP realigns gas costs with actual computational resource consumption, based on benchmarking data rather than historical convention. Gas costs for certain operations are being recalibrated to reflect modern hardware performance, which should reduce effective fees for common smart contract interactions.
EIP-8037 and EIP-8038 increase gas costs for state creation and state access respectively, discouraging state bloat — a persistent problem where Ethereum's state trie grows unchecked, increasing node hardware requirements.
EIP-7954 raises the maximum contract bytecode size, accommodating more complex on-chain applications without requiring proxy patterns or code-splitting workarounds.
Over 25 non-headliner EIPs remain under consideration. Developers explicitly declined to include EIP-7805 (FOCIL — Fork-Choice Inclusion Lists), reduced slot times, multidimensional gas metering, and post-quantum signature verification in this fork. Base's engineering team publicly warned that adding FOCIL alongside ePBS could delay the upgrade beyond 2026.
As of late March 2026, Glamsterdam has entered Devnet-5 testing. The All Core Devs Execution (ACDE) call #233 on March 26 included Glamsterdam devnet updates and benchmarking presentations. Development proceeds on two parallel devnet tracks:
Three of the proposed EIPs have cleared Devnet-4 testing. The transition to Devnet-5 focuses on cross-client compatibility — ensuring Geth, Besu, Prysm, and other execution and consensus clients communicate correctly under the new payload separation model.
Public testnets and dual audit phases are outlined for spring 2026, creating the earliest realistic path toward a June activation. However, core developers have been explicit that the date is secondary to correctness. Given the structural complexity of ePBS — which rewrites the fundamental validator-builder interaction at the consensus layer — a slip into Q3 or Q4 2026 remains plausible.
The precedent is not reassuring. The Merge was delayed multiple times over two years. Pectra and Fusaka, Glamsterdam's predecessors in 2025, both shipped later than initial targets.
For validators: ePBS lowers the technical barrier to solo validation. Currently, competitive validators must integrate with MEV-Boost relays and maintain infrastructure to receive builder payloads. Under ePBS, a validator's consensus-layer duties are cleanly separated from execution-layer complexity. This could improve the economics of solo staking by removing the need for sophisticated MEV infrastructure, though the actual revenue impact depends on whether ePBS auctions produce comparable yields to current relay-based auctions.
For builders: The competitive dynamics intensify. Builders must now stake ETH on the beacon chain, introducing capital costs. Those who fail to deliver payloads after committing face slashing. The sealed-bid mechanism removes information asymmetries that currently allow certain relay operators to preference specific builders. The market may consolidate further around builders with the deepest capital reserves and most exclusive order flow, or it may open slightly as the permissionless on-chain auction removes relay-level gatekeeping.
For MEV: Total extractable value does not change. ePBS restructures how MEV is competed for, not whether it exists. Sandwich attacks, arbitrage, and liquidation extraction continue. The economic difference is in distribution — more of the MEV surplus should flow to proposers (validators) through competitive bidding, rather than being captured by builders through exclusive relay relationships.
For the broader network: The gas limit increase from 60 million to 200 million per block, phased alongside BALs, targets approximately 10x throughput improvement. Combined with gas repricing (EIP-7904), the effective cost of common operations should decrease. Current average gas prices of 0.09 gwei — already at historic lows following the Dencun upgrade — may not change dramatically, but throughput at that price level expands substantially.
Ethereum developers have already named the post-Glamsterdam upgrade "Hegota," slated for late 2026. Its confirmed headliner is FOCIL (EIP-7805): Fork-Choice Obligatory Commitment to Inclusion Lists.
Under FOCIL, 16 randomly selected attesters each nominate transactions for mandatory inclusion in the next block. If a block is missing nominated transactions without justification, it can be rejected by the network. This mechanism directly targets censorship resistance — even a dominant builder cannot exclude transactions that FOCIL attesters have nominated.
Buterin has outlined a further evolution: "Big FOCIL," which would narrow builders' role to MEV ordering and execution optimization while inclusion decisions are distributed across the attester committee. Combined with encrypted mempools — where transaction contents are hidden until after block commitment — this would eliminate most forms of frontrunning and sandwich attacks at the protocol level.
The roadmap beyond Hegota includes network-layer privacy (Tor routing, Ethereum-specific mixnets), ZK-EVM integration targeting 1,000x throughput improvements, and what Buterin described as a BitTorrent-like scaling model that relaxes Ethereum's synchronous shared-state assumption for certain transaction categories.
Glamsterdam is a containment upgrade, not a decentralization solution. It accepts that block building is concentrated and builds protocol-level firewalls to prevent that concentration from compromising Ethereum's consensus layer. The sealed-bid auction mechanism, builder staking, and elimination of relay dependencies represent the most significant architectural change to Ethereum's block production since MEV-Boost launched in 2022.
The economic question is whether ePBS improves value capture for validators and solo stakers sufficiently to reverse the ongoing decline in solo staking participation, or whether the capital requirements of builder staking further entrench incumbents. The technical question is whether the Ethereum Foundation can ship a consensus-layer rewrite of this magnitude on schedule, given the project's historical pattern of delays.
What is not in question is the direction. Ethereum's 2026 roadmap — Glamsterdam, then Hegota, then encrypted mempools — represents a multi-year commitment to restructuring the block production pipeline from the protocol layer up. Whether the timeline holds is uncertain. That the work is necessary is not.