Ethereum's Glamsterdam hard fork — the network's largest protocol change since The Merge — has slipped from its original June 2026 target to Q3 2026. The upgrade centers on two confirmed EIPs: EIP-7732 (enshrined Proposer-Builder Separation, or ePBS) and EIP-7928 (Block-Level Access Lists, or BAL...
"There's a new chapter starting for the Protocol cluster." — Will Corcoran, Ethereum Foundation Protocol Cluster Lead, May 11, 2026
Ethereum's Glamsterdam hard fork — the network's largest protocol change since The Merge — has slipped from its original June 2026 target to Q3 2026. The upgrade centers on two confirmed EIPs: EIP-7732 (enshrined Proposer-Builder Separation, or ePBS) and EIP-7928 (Block-Level Access Lists, or BALs). Together, they would raise Ethereum's gas limit from approximately 60 million to 200 million, enable parallel transaction execution across CPU cores, and reduce gas fees by an estimated 78.6% across both simple transfers and complex smart contract calls.
The stakes are structural. Approximately 93% of Ethereum blocks currently route through MEV-Boost, an off-chain relay system where four operators control 95% of market share. ePBS would eliminate dependency on these trusted intermediaries by enshrining builder-proposer separation directly into Ethereum's consensus layer. The upgrade arrives against a backdrop of institutional uncertainty: eight senior Ethereum Foundation researchers have departed in 2026, five of them in May alone, including all three protocol leads. Whether the reconstituted team can deliver Glamsterdam on the revised Q3 timeline is the central open question.
Glamsterdam is a coordinated hard fork across Ethereum's execution and consensus layers. According to the Ethereum Foundation's May 2, 2026 Soldøgn Interop recap, the upgrade locks in a post-fork gas limit floor of 200 million — a 3.3x increase from the current ~60 million ceiling. The performance target: 10,000 transactions per second on Layer 1, with a projected 78.6% reduction in gas costs per operation.
For context, Ethereum L1 currently processes approximately 15-30 TPS. Layer 2 networks handle the bulk of throughput, accounting for roughly 95% of Ethereum's total transaction volume according to 2026 on-chain data. Glamsterdam does not replace L2 scaling; it expands L1 headroom so rollups, blob data, and base-layer DeFi can coexist without contention.
The upgrade originally targeted June 2026. On May 11, the Ethereum Foundation confirmed a slip to Q3, citing scope complexity. According to QuickNode's technical breakdown from March 2026, over 25 non-headliner EIPs were under consideration for inclusion — a volume that community members described internally as "an absolute beast."
This is the centerpiece. Under the current MEV-Boost system, block construction is outsourced to off-chain builders through trusted relay intermediaries. EIP-7732 moves this entire mechanism on-chain.
The mechanics: builders assemble blocks and cryptographically seal their contents, publishing bids with payload commitments. Proposers (validators) select the highest-paying block without seeing or tampering with its contents. Transactions are only revealed after the block is finalized. This eliminates the need for external relay operators.
According to Flashbots documentation, MEV-Boost currently handles approximately 93% of Ethereum block production. Just four relay operators control 95% of the relay market, with bloXroute operating two relays and controlling nearly 30% of market share alone. Two to three builders produce 80-95% of all blocks. ePBS is designed to reduce this concentration by removing the trust assumptions that entrench incumbents.
However, the change carries its own centralization risk. According to research simulations cited in academic literature, ePBS could increase profit concentration among a small number of builders who win auctions consistently, potentially reinforcing — rather than resolving — builder-level centralization.
BALs are structured declarations specifying exactly which storage slots and accounts each transaction will read or write. A hash of this access list — the "BAL root" — is embedded directly in the block header.
The purpose: parallel execution. By pre-declaring state access patterns, Ethereum moves from sequential processing (one transaction at a time) to a multi-lane system where non-conflicting transactions execute simultaneously across multiple CPU cores. This is the primary mechanism enabling the 3.3x gas limit increase without proportionally increasing block processing time.
EL clients must store BALs for a minimum of 3,533 epochs. The upgrade introduces a new block header field (block_access_list_root) and a new peer-to-peer protocol version (eth/71) for BAL distribution across the network.
According to QuickNode's analysis, ePBS and BALs together are expected to deliver 10-30x improvements in execution throughput short-term. ZK-EVMs targeting longer-term gains of up to 1,000x remain separate from Glamsterdam's scope.
Beyond the two headliners, the Soldøgn Interop confirmed several additional EIPs for inclusion:
| EIP | Description | Status |
|-----|-------------|--------|
| 7904 | General gas repricing — aligns gas costs with actual computational resource consumption | Included |
| 8037 | State creation gas cost increase — raises cost of new state storage operations | Finalized (moved from dynamic to fixed cost_per_state_byte model) |
| 8038 | State-access gas cost increase | Under consideration |
| 8061 | Exit/consolidation churn limit tied to total staked ETH | Included |
| 7954 | Increase maximum contract size | Under consideration |
| 8045 | Slashed-validator duty removal (scoped to proposer duties only) | Modified and included |
| 7688 | SSZ stable containers | In scope but deferred from initial devnet |
EIP-7904 is notable because it recalibrates gas prices that, according to the Ethereum Foundation, were set years ago and no longer reflect execution costs on modern hardware. The repricing covers compute, state operations, data handling, and gas accounting — a comprehensive realignment.
EIP-8061 addresses staking liquidity. Based on approximately 38.9 million ETH currently staked, the revised formula produces an exit churn limit of roughly 1,187 ETH per epoch — approximately 4.6x larger than the current limit. This directly benefits institutional stakers managing large positions, as exit queues currently constrain their ability to rebalance or de-risk.
The economic rationale for ePBS is rooted in measurable centralization. Current data from The Block and relay monitoring services:
This concentration emerged because relay operation is economically unviable as a standalone business. Blocknative shut down its MEV-Boost relay service due to the absence of a reasonable fee mechanism. The market consolidated around operators willing to absorb losses for strategic positioning.
ePBS would eliminate relays entirely, replacing trust-based off-chain coordination with protocol-enforced auction mechanics. Builders still compete, but the infrastructure layer becomes part of Ethereum's consensus rather than a third-party dependency.
The counterargument, documented in academic research: enshrining the auction on-chain may give well-capitalized builders a structural advantage in winning blocks, potentially making builder-level centralization harder to displace since it would be embedded in protocol rules rather than in removable middleware.
The Soldøgn Interop took place April 28 – May 2, 2026, in Longyearbyen, Svalbard — a location at 78°N latitude, above the Arctic Circle. Just over 100 Ethereum core contributors attended. According to the official Ethereum Foundation recap authored by Tim Beiko:
Monday: Initial devnet attempt with 4 execution layer × 4 consensus layer client configurations failed; pushed to Tuesday.
Tuesday: A 4×3 configuration achieved stability. Stress testing began. Builder API breakout sessions simplified the ePBS spec.
Wednesday: Edge cases in reservoir refund/refill mechanics surfaced. EIP-8237 (top-up sync architecture) was deferred to a future fork.
Thursday: CL teams reported stable ePBS. EL bid pathways were still debugging. Design decisions on FOCIL (Fork-Choice Inclusion Lists) were finalized — specifically, the decision to exclude FOCIL from Glamsterdam.
Friday: Nearly all clients ran together with end-to-end external builder pipelines tested. The post-Glamsterdam gas limit floor of 200 million was locked in.
The glamsterdam-devnet-2 was running stable multi-client implementations with external builders by week's end. EthPandaOps delivered an ethIQ performance dashboard and a panda MCP server for agentic development workflows.
Glamsterdam's delivery timeline intersects with significant institutional upheaval at the Ethereum Foundation. According to reporting from CoinDesk (May 18 and May 21, 2026), Unchained, and CryptoPotato:
The departures followed the EF's March 13, 2026 publication of "The Promise of Ethereum: Introducing the EF Mandate," which established the CROPS framework — Censorship Resistance, Open source, Privacy, and Security — as core organizational principles. Reports from Cryptopolitan indicated that all EF members were asked to sign the Mandate document, though this claim was unconfirmed by EF communications.
New leadership has been named. Will Corcoran assumed the Protocol Cluster lead role, with Kev Wedderburn and Fredrik as additional leads. Pari is serving as interim ACDC moderator, and Barnabas is covering ACDT. Nixo and Ansgar remain as ACDE co-chairs.
Dankrad Feist, a former EF researcher, stated publicly: "The way to save Ethereum is for the community to create an organization that's economically aligned with Ethereum and accountable to it." He noted that the EF controls less than 0.1% of all ETH and receives no direct staking or fee revenue — a structural misalignment between the organization responsible for protocol development and the network's economic value.
Two significant features were explicitly excluded from Glamsterdam:
FOCIL (Fork-Choice Inclusion Lists, EIP-7805): This censorship-resistance mechanism would allow multiple validators — rather than a single block builder — to enforce transaction inclusion through Ethereum's fork-choice rule. It received "exceptionally strong support" from the community but was moved to Hegotá, the subsequent upgrade planned for late 2026.
The Base engineering team publicly warned that shipping FOCIL alongside ePBS could delay Glamsterdam beyond 2026. According to a Base blog post, the team supports FOCIL's censorship resistance properties but opposed introducing additional complexity into an already dense upgrade. Early FOCIL prototypes were functional at the Soldøgn Interop, with multi-client interop planned for the Hegotá development cycle.
Verkle Trees: Originally planned for Glamsterdam, Verkle Trees would cut node storage requirements by up to 90% and enable stateless clients — nodes that can validate blocks without storing the full state. This feature has been moved to Hegotá, transforming it into what community members describe as a "cleanup and hardening" fork.
The Hegotá upgrade is also expected to include account abstraction improvements (covering signature schemes, aggregation, batching, recovery, and gas sponsorship) and a QUIC-based libp2p replacement. An erasure-coded broadcast prototype demonstrated at Soldøgn showed approximately 6x faster propagation than GossipSub on 2.4 MB payloads.
Glamsterdam's economic impact operates on several layers:
Fee reduction: The 78.6% gas cost reduction, if realized, directly compresses L1 transaction costs. EIP-7904's repricing recalibrates costs that have been static for years despite hardware improvements. For DeFi protocols operating on L1 Ethereum, this represents a material reduction in execution costs.
L2 capacity: The 200 million gas limit expands L1 headroom for rollup settlement and blob data. Layer 2 networks currently process approximately 5,600 TPS collectively. Developer projections cited in industry analysis suggest capacity could reach 24,000+ TPS as Glamsterdam's changes propagate. Industry estimates suggest blob fees could contribute 30-50% of total ETH burn by 2026, depending on L2 activity levels.
Staking economics: EIP-8061's 4.6x increase in the exit churn limit improves capital efficiency for institutional stakers. With 38.9 million ETH staked, the exit queue bottleneck has been a friction point for large-position managers. Faster exit capacity reduces the liquidity risk premium demanded by institutional allocators.
ETH burn dynamics: Higher throughput at lower per-transaction costs could increase total gas consumed even as individual costs fall. Whether this nets positive or negative for ETH supply via EIP-1559 burning depends on realized demand elasticity — a variable that remains empirically unresolved.
Glamsterdam represents Ethereum's most ambitious protocol change since The Merge. The technical scope is well-defined: ePBS removes a centralization bottleneck, BALs unlock parallelism, and gas repricing brings fee structures into alignment with modern hardware. Multi-client devnets are running, and the core specifications are frozen.
The execution risk is institutional, not technical. The Ethereum Foundation's Protocol Cluster has lost its most experienced contributors at the precise moment Glamsterdam demands coordination across every major client team. Will Corcoran's reconstituted leadership must shepherd a Q3 mainnet activation while simultaneously scoping Hegotá, onboarding replacements, and managing a community that has not received a public explanation for the departures.
The Soldøgn Interop demonstrated that the engineering talent across client teams remains capable — devnet stability was achieved in days, not weeks. But the coordination layer above client teams, the function the EF Protocol Cluster historically performed, is operating with new personnel and untested relationships. Whether that layer holds through a Q3 mainnet fork will determine Glamsterdam's outcome more than any EIP specification.