Ethereum's next hard fork, Glamsterdam, will enshrine Proposer-Builder Separation (ePBS) directly into the protocol via EIP-7732, eliminating the network's dependency on MEV-Boost middleware that currently intermediates over 90% of block production. The upgrade, originally targeting June 2026, ha...
Ethereum's next hard fork, Glamsterdam, will enshrine Proposer-Builder Separation (ePBS) directly into the protocol via EIP-7732, eliminating the network's dependency on MEV-Boost middleware that currently intermediates over 90% of block production. The upgrade, originally targeting June 2026, has shifted to Q3 2026 following scope negotiations at the Soldøgn Interop in Svalbard, Norway, where over 100 core contributors confirmed three milestones: a post-upgrade gas limit floor of 200 million (3.3x the current 60 million), stable multi-client ePBS devnets with external builder pipelines, and finalized EIP-8037 state-access repricing parameters.
The stakes are structural. Today, two builders—Titan (44.46% market share) and Beaverbuild (42.53%)—produce approximately 87% of all Ethereum mainnet blocks. Seven relay operators control 99% of the MEV-Boost network. Private order flows contribute to 60% of MEV in over half of all blocks. Glamsterdam's ePBS removes the relay layer entirely, replacing it with a trustless, protocol-native auction that researchers estimate could reduce MEV extraction by up to 70%.
MEV-Boost launched after Ethereum's Merge in September 2022 as an out-of-protocol mechanism to democratize block construction. By 2026, it has produced the opposite result.
Current block builder market concentration (May 2026):
| Builder | Market Share | |---------|-------------| | Titan Builder | 44.46% | | Beaverbuild | 42.53% | | Rsync | 9.98% | | All others | ~3.03% |
The Herfindahl-Hirschman Index (HHI) for Ethereum's block builder market stands at approximately 3,892—more than double the U.S. Department of Justice threshold of 1,800 for a "highly concentrated" market.
The relay layer adds another centralization vector. Seven unique relay operators control 99% of MEV-Boost traffic. Private order flow—transactions sent directly to builders rather than the public mempool—accounts for 60% of MEV in more than half of all blocks. New builder entrants face subsidy requirements of 1.4 ETH or more simply to establish credibility for order flow access, creating an implicit barrier to competition.
Over 90% of Ethereum blocks are currently produced through the MEV-Boost auction system. This means Ethereum's consensus layer runs a trustless, decentralized validator set, but its execution layer depends almost entirely on a permissioned, centralized supply chain.
Enshrined Proposer-Builder Separation formally separates the proposer role (selecting the consensus block) from the builder role (assembling the execution payload) at the protocol level. The mechanism:
The critical innovation is the Payload Timeliness Committee (PTC)—a randomly-selected subset of 512 validators per slot whose sole function is to attest whether the execution payload arrived on time.
PTC dual-deadline mechanism:
PTC members do not execute the block. They verify arrival timing and commitment matching only.
This architecture expands the data propagation window from the current ~2 seconds to approximately 9 seconds, directly enabling higher throughput. The relay layer—currently intermediating trust between proposers and builders—becomes unnecessary because the protocol itself enforces the trustless exchange of payload for payment.
Ethereum's gas limit has followed a deliberate scaling path:
The 200M target represents a 3.3x capacity increase from the current 60M limit. This is technically feasible under ePBS because the restructured slot architecture gives execution clients substantially more processing time.
The mechanism:
According to analysis cited by The Defiant, the combination could triple Ethereum's execution capacity, with some estimates reaching 10,000 TPS at the 200M gas limit under favorable conditions.
A 3.3x gas limit increase without guard rails would accelerate Ethereum's state growth unsustainably. EIP-8037 addresses this with a state-creation gas cost increase.
Key parameters (finalized at Soldøgn):
EIP-8037 introduces a state gas reservoir model that splits accounting between computational gas (processing) and state gas (permanent storage). State gas charges draw from the reservoir first, preventing execution frames from miscalculating available gas.
The earlier dynamic per-state-byte model was dropped during Soldøgn in favor of a fixed cost_per_state_byte approach, simplifying implementation and predictability.
From April 28 to May 2, 2026, over 100 Ethereum core contributors convened in Longyearbyen, Svalbard (78°N latitude) for the Soldøgn Interop—named after the Arctic midnight sun.
Goals and outcomes:
| Goal | Status | |------|--------| | Post-Glamsterdam gas limit floor of 200M | Aligned | | Stable ePBS with external builder pipeline | Achieved (multi-client) | | EIP-8037 repricing parameters finalized | Locked |
Teams began the week targeting a 4 EL × 4 CL (Execution Layer × Consensus Layer) Glamsterdam devnet by Monday evening. Initial implementation issues pushed delivery to Tuesday, when a 4×3 configuration ran stably enough for stress testing to commence.
By Friday, nearly all clients were running together on glamsterdam-devnet-2 with the external builder pipeline tested end-to-end. Progress was also made on Hegotá features (FOCIL, native account abstraction) that will follow Glamsterdam.
Three Ethereum Foundation teams provided infrastructure: EthPandaOps deployed ethIQ and a panda MCP server for agentic developer workflows; Protocol Support ran soldogn.xyz as the coordination hub; and the EF Digital Studio team documented proceedings.
Current timeline:
Key risk: scope creep
The Base engineering team publicly warned that adding FOCIL (Fork-Choice Inclusion Lists) alongside ePBS could delay the upgrade beyond 2026. In response, Ethereum core developers moved FOCIL to the subsequent Hegotá upgrade, keeping Glamsterdam's scope manageable. Hegotá will also introduce Verkle Trees (targeting 90% node storage reduction).
Over 25 non-headliner EIPs remain under consideration for inclusion, each representing potential integration complexity and testing overhead.
Client readiness: The Soldøgn devnet achieved a 4×3 (later near-full) multi-client configuration. Production readiness requires completing hardening, test coverage, and merging the draft PRs produced during the interop.
For validators: ePBS formalizes the proposer's revenue model. Proposers receive builder payments directly through the protocol rather than through trusted relay intermediation. This removes counterparty risk but may compress margins as the builder auction becomes more transparent.
For builders: The elimination of relay trust requirements lowers barriers for new entrants. However, access to private order flow—the dominant competitive moat—remains an off-protocol advantage. Whether ePBS meaningfully deconcentrates the builder market depends on how private order flow dynamics evolve.
For users: Researchers estimate up to 70% reduction in MEV extraction. At the current ~$500M annual MEV extracted from Ethereum users (per Flashbots data), this implies up to $350M in annual savings returned to transaction originators. The 200M gas limit, combined with parallel execution via BALs, should reduce base fees under normal load conditions.
For L2s: More blobs per block (enabled by the wider propagation window) further reduces L2 data availability costs, compounding the 40–60% fee reduction already delivered by Fusaka's PeerDAS.
Value distribution shift: Glamsterdam moves economic rents from relay operators and incumbent builders toward validators (who capture builder payments directly) and users (who face less MEV extraction). This is consistent with the economic value framework's observation that middleware intermediaries capture outsized rents in opaque supply chains.
Glamsterdam represents Ethereum's most consequential architectural change since the Merge. It addresses the network's central contradiction: a decentralized consensus layer running atop a centralized execution supply chain. By enshrining proposer-builder separation into the protocol and simultaneously unlocking a 3.3x gas limit increase, the upgrade attempts to resolve both the fairness problem (MEV extraction) and the throughput problem (limited execution capacity) in a single coordinated deployment.
The Q3 2026 timeline is aggressive given the multi-client coordination required. The Soldøgn Interop demonstrated technical feasibility—4×3 devnets running with external builders—but production hardening remains ahead. The decision to defer FOCIL to Hegotá was a pragmatic scope reduction that trades short-term censorship resistance improvements for timely delivery of the structural reforms.
Whether ePBS achieves its decentralization goals depends on factors outside the protocol's control—particularly private order flow dynamics. The relay layer will be eliminated, but if two builders still control 87% of blocks because of exclusive deal flow, the concentration problem migrates rather than resolves. This will be the key metric to track post-deployment.