Ethereum's Glamsterdam upgrade — the network's largest protocol overhaul since the 2022 Merge — entered its final devnet phase in mid-June 2026, with all planned Ethereum Improvement Proposals (EIPs) now running together in multi-client developer networks. The upgrade combines Enshrined Proposer-...
"This is probably the largest fork we've had since the Merge. It will change a lot of assumptions about Ethereum and set us up for much more scaling in the future." — Parithosh Jayanthi, Ethereum Foundation Core Developer
Ethereum's Glamsterdam upgrade — the network's largest protocol overhaul since the 2022 Merge — entered its final devnet phase in mid-June 2026, with all planned Ethereum Improvement Proposals (EIPs) now running together in multi-client developer networks. The upgrade combines Enshrined Proposer-Builder Separation (ePBS) under EIP-7732 with Block-Level Access Lists (BALs) under EIP-7928 to unlock parallel transaction processing and a tripling of the network's gas limit from approximately 60 million to a 200 million floor.
Originally targeted for June 2026, the timeline has slipped to Q3 2026 at the earliest. According to the Ethereum Foundation, public testnets on Holesky and Hoodi must hold stable across multiple epochs before mainnet activation, placing the realistic window between September and December 2026. The delay coincides with leadership turnover at the Foundation's protocol cluster: Barnabé Monnot and Tim Beiko are departing, Alex Stokes is taking sabbatical, and Will Corcoran now leads the team alongside Kev Wedderburn.
The economic implications are material. Glamsterdam's gas repricing restructures Ethereum's fee market — making high-level compute operations cheaper while increasing costs for state-access operations — and the ePBS integration removes the network's current 93% dependency on off-protocol MEV-Boost relays for block construction. For L2 rollups that already account for 77% of Ethereum's $48 billion execution market, the upgrade narrows L1's cost and throughput disadvantages but simultaneously reduces the fee pressure that burns ETH on the base layer.
Glamsterdam bundles at least 12 EIPs across four categories: L1 scaling, network sustainability, network resilience, and developer experience. The upgrade is formally designated EIP-7773 and combines the Amsterdam execution-layer fork with the Gloas consensus-layer fork.
The two headline proposals are:
EIP-7732 (Enshrined Proposer-Builder Separation): Moves the block-building process from off-protocol relay infrastructure directly into Ethereum's consensus layer. Builders assemble blocks and publish cryptographically sealed payload commitments. Proposers select the highest-paying block without visibility into its contents. The data-propagation window expands from approximately 2 seconds to 9 seconds, and a new Payload Timeliness Committee validates execution.
EIP-7928 (Block-Level Access Lists): Attaches an upfront map to each block listing accounts and storage slots that transactions will access. This dependency mapping enables nodes to identify non-conflicting transactions and distribute them across multi-core CPUs for parallel execution rather than processing them serially.
Supporting EIPs include EIP-8159 (networking protocol for sharing access lists), EIP-8037 (fixed cost-per-state-byte targeting 120 GiB/year state growth), EIP-8038 (increased gas costs for EXTCODESIZE and EXTCODECOPY), EIP-2780 (reducing intrinsic transaction gas for up to 71% cheaper ETH transfers), EIP-7997 (deterministic factory predeploy for identical contract addresses across EVM chains), and EIP-8045 (excluding slashed validators from proposing).
EIP-8080 enables exits to use the consolidation queue, allowing up to 2.5x faster exit speeds during high-demand periods — a direct response to queuing issues observed after the Pectra upgrade in May 2025.
The MEV-Boost relay layer represents one of Ethereum's most concentrated infrastructure dependencies. According to data from The Block, 93% of Ethereum blocks created in the most recent 14-day period used MEV-Boost relays. Only four relay operators remain active on the network, and five entities relay 98% of all MEV-boosted transactions.
This concentration occurred partly through attrition. Blocknative discontinued its relay in 2023 because operating one was not economically viable — relayers collect no fees for their services. The result is that a critical piece of Ethereum's block production infrastructure operates as an unprofitable public good maintained by a shrinking number of entities.
ePBS eliminates this dependency by integrating builder-proposer handoff logic directly into the consensus layer. Under the new design, builders compete through protocol-native mechanisms rather than trusting off-chain relay infrastructure. MEV-Boost relays become optional rather than required. According to ethereum.org documentation, the protocol will natively settle builder payments, removing the trust assumptions that currently make relays necessary.
The practical effect: block construction moves from a system where 93% of blocks route through four off-protocol operators to one where the protocol itself mediates the builder-proposer relationship. This does not eliminate MEV extraction — builders will still optimize transaction ordering for profit — but it removes the centralized relay bottleneck through which that extraction currently flows.
The Ethereum Foundation established a 200 million gas-limit floor as the post-Glamsterdam target, confirmed at an interop event in Svalbard, Norway. This represents roughly 3.3x the current approximately 60 million gas limit. The upgrade does not force this limit; validators set gas limits through standard gas-vote signaling, which they currently coordinate around the 60 million range. Glamsterdam enables the higher limit by providing the parallel-execution infrastructure needed to process blocks of that size.
The throughput design target is approximately 10,000 transactions per second, roughly 10x current capacity. However, as CryptoBriefing noted in its June 17, 2026 coverage, "actual throughput gains depend on multiple factors: client performance, validator hardware, gas repricing, and how much parallelism real blocks can actually support." The 10,000 TPS figure is a theoretical ceiling, not a guarantee.
Testing at a 150 million reference block gas limit is underway in current devnets, with the full 200 million floor to be validated on public testnets. Additional EIPs remain under active testing in devnets, including EIP-7778, EIP-7843, EIP-7976, EIP-7981, and EIP-8024.
Glamsterdam restructures Ethereum's gas cost tables. According to CoinDesk reporting from June 16, 2026, the repricing makes high-level compute operations cheaper while state-access operations become more expensive. The goal is to align fee structures with actual resource consumption.
EIP-2780 reduces intrinsic transaction gas, cutting standard ETH transfer costs by up to 71%. EIP-8037 introduces a fixed cost-per-state-byte with a dedicated gas reservoir for state growth, targeting a maximum state expansion of 120 GiB per year. EIP-7954 raises the maximum smart contract size from approximately 24 KiB to 32 KiB.
The winners: applications that perform heavy computation but minimal state writes. DeFi protocols running complex math (pricing engines, risk models) benefit from cheaper compute. Users making simple ETH transfers see the largest percentage reduction.
The losers: applications that create substantial new on-chain state. Protocols that deploy many new contracts or write extensively to storage face higher costs under the new pricing. The repricing also facilitates zero-knowledge proving systems by making verification operations less expensive relative to state creation — aligning economic incentives with Ethereum's ZK-rollup scaling strategy.
Glamsterdam's development timeline, as of July 1, 2026:
| Phase | Status | Notes | |-------|--------|-------| | EIP finalization | Complete | 12+ EIPs included in EIP-7773 | | ePBS stabilization | Complete | Confirmed at Svalbard interop | | EIP-8037 finalization | Complete | State-pricing model locked | | Final devnets (Devnet-5) | In progress | Multi-client, full EIP suite since mid-June | | Public testnets (Holesky, Hoodi) | Pending | Must hold stable across multiple epochs | | Client releases & security audits | Pending | Follows testnet stability | | Mainnet activation | H2 2026 target | September–December 2026 realistic window |
The Ethereum Foundation characterized the current phase as "the last step before hardening and shipping to public testnets." Given that recent Ethereum forks have required two to four months of public-testnet seasoning, a September activation represents an optimistic case, with November–December more likely if any devnet issues surface.
Ethereum Foundation developers noted that Glamsterdam is "proving trickier and slower than Fusaka," the prior upgrade cycle, suggesting that the complexity of integrating ePBS and parallel execution simultaneously has introduced engineering challenges beyond initial estimates.
Glamsterdam's final development phase coincides with the most significant personnel turnover at the Ethereum Foundation's protocol team in years. Tim Beiko, who coordinated Ethereum's All Core Developers calls and managed the upgrade process through multiple hard forks, is departing. Barnabé Monnot, a protocol researcher focused on MEV and proposer-builder dynamics, is also leaving. Alex Stokes, who contributed to consensus-layer research, is taking sabbatical.
Will Corcoran, who now leads the protocol cluster, stated: "There's a new chapter starting for the Protocol cluster." He is joined by Kev Wedderburn and developer Fredrik. The transition occurs as the Foundation also advances work on the subsequent Hegotá upgrade (targeting late 2026 as a "cleanup and hardening" fork) and the longer-horizon quantum-resistant Strawmap roadmap.
The departures raise continuity questions. Beiko's institutional knowledge of Ethereum's upgrade process and developer coordination, and Monnot's expertise in MEV economics — the exact domain ePBS addresses — leave gaps that the new leadership must fill during Glamsterdam's most critical testing phase.
Glamsterdam recalibrates the economic relationship between Ethereum L1 and its L2 ecosystem. Two L2 networks currently hold 77% of Ethereum's approximately $48 billion rollup market, according to webthreepedia's prior coverage. The Pectra upgrade in May 2025 increased blob throughput, making data availability cheaper for rollups.
Glamsterdam adds another dimension: L1 throughput rising toward 10,000 TPS and transfer costs falling by up to 71% reduce the performance gap that drove applications to L2 in the first place. For simple transactions and smaller DeFi operations, L1 may become cost-competitive with rollups for the first time since EIP-4844 made L2 settlement cheap.
The tension is structural. More L1 capacity and cheaper L1 gas means some activity that currently settles on L2 may migrate back to L1. However, L2s that have built ecosystems, user bases, and sequencer revenue may retain activity through network effects rather than cost advantages alone.
For ETH's monetary economics, the dynamic is ambiguous. Cheaper gas on L1 could increase transaction volume (more activity, more fee burns) or decrease per-transaction revenue (cheaper fees, less burn per transaction). The net effect on ETH supply reduction through EIP-1559 burning depends on elasticity of demand — a variable that remains empirically untested at the throughput levels Glamsterdam targets.
Glamsterdam represents the most extensive single modification to Ethereum's protocol architecture since The Merge. The upgrade attacks two systemic vulnerabilities simultaneously: the centralized relay infrastructure that mediates 93% of block production and the serial execution model that caps L1 throughput at a fraction of competing chains. Whether the engineering complexity of integrating ePBS and parallel execution in a single fork proves manageable within the H2 2026 window remains the primary risk. The Foundation's own developers acknowledge Glamsterdam is proving more difficult than expected, and the concurrent leadership transition adds organizational uncertainty to a technically demanding schedule. The data will emerge from public testnets in the coming months.