Ethereum's next hard fork, Glamsterdam, entered its first generalized devnet phase in late April 2026 after months of component-level testing. The upgrade bundles two structural changes — enshrined Proposer-Builder Separation (ePBS, EIP-7732) and Block-Level Access Lists (BALs, EIP-7928) — that t...
"Core devs are targeting next week for the 1st Glamsterdam devnet for Ethereum's next upgrade. ePBS has been an incredibly structurally complex change, splitting block production into 2 coordinating parties inside consensus." — Nixo, Ethereum Core Developer, April 2026
Ethereum's next hard fork, Glamsterdam, entered its first generalized devnet phase in late April 2026 after months of component-level testing. The upgrade bundles two structural changes — enshrined Proposer-Builder Separation (ePBS, EIP-7732) and Block-Level Access Lists (BALs, EIP-7928) — that together target a 10x increase in Layer 1 throughput and a 78.6% reduction in gas fees. The gas limit is set to rise from 60 million to 200 million per block.
Developer documentation references June 2026 as the aspirational mainnet target. Core teams have not committed to a date. The Ethereum Foundation's April 2026 Checkpoint #9 report characterized the timeline as "unlikely" for Q2, pointing to ePBS implementation complexity as the primary bottleneck. A Q3 or Q4 2026 deployment is considered more realistic by multiple development teams.
At stake is Ethereum's competitive positioning. The network currently processes 15–30 TPS on Layer 1 while 89% of ecosystem transactions run through Layer 2 rollups. Glamsterdam's parallel execution capability targets approximately 10,000 TPS — a figure that would place Ethereum L1 in direct competition with Solana's observed 1,000–1,500 TPS and Monad's 10,000+ TPS claims. ETH trades near $2,250 as of May 1, 2026.
Glamsterdam is a portmanteau of "Gloas" (the star) and "Amsterdam" (the host city of Devconnect 2022). It follows the Fusaka upgrade, which went live in December 2025 and introduced PeerDAS for data availability sampling alongside a gas limit increase to 60 million.
Vitalik Buterin outlined eight EIPs for Glamsterdam in late February 2026, organized across five categories: L1 scaling and parallel processing, network sustainability, network resilience, user and developer experience, and gas repricing. The full scope, according to the ethereum.org roadmap page, includes:
Scaling and Execution:
Gas Economics:
User Experience:
Network Resilience:
Over 30 proposals were declined for inclusion, deferred to the subsequent Hegotá upgrade.
The headline consensus-layer change is EIP-7732, which moves block building from off-chain relay infrastructure directly into the Ethereum protocol. The economic rationale is straightforward: approximately 80–90% of Ethereum block production currently depends on off-chain services, primarily MEV-Boost relays operated by a small number of entities.
According to Relayscan data from April 2026, the relay market breaks down as follows:
| Relay | Market Share (24h) | |-------|-------------------| | relay.ultrasound.money | 33.9% | | titanrelay.xyz | 24.2% | | bloxroute (max-profit) | 14.7% | | aestus.live | 10.0% | | bloxroute (regulated) | 9.1% | | boost-relay.flashbots.net | 4.2% |
On the builder side, concentration is more acute. Titan controls 52.2% of blocks, BuilderNet handles 24.6%, and Quasar accounts for 15.1%. Two to three builders effectively control the vast majority of block production.
ePBS addresses this by integrating the proposer-builder separation mechanism into the consensus layer itself. The protocol introduces a Payload Timeliness Committee (PTC) and dual-deadline logic, expanding the propagation window from 2 seconds to approximately 9 seconds. Validators propose block headers; builders fill them. The trust assumption shifts from relay operators to the protocol.
A secondary concern, flagged by Buterin in a March 2026 post, is that ePBS could inadvertently concentrate power among sophisticated block builders. The proposed countermeasure — FOCIL (Forward Obligatory Commitment to Inclusion Lists, EIP-7805) — was not included in Glamsterdam but was selected as the consensus-layer headliner for the follow-on Hegotá upgrade.
Approximately 30% of current blocks comply with OFAC sanctions lists through relay-level filtering, enabling systematic transaction exclusion. ePBS does not directly resolve censorship resistance at the builder level, but it removes the relay as a single chokepoint for filtering decisions.
EIP-7928 introduces Block-Level Access Lists (BALs), which require each block to pre-declare which accounts and storage slots its transactions will access. This provides an upfront dependency map that allows the execution layer to process non-conflicting transactions across multiple CPU cores simultaneously, rather than sequentially.
The estimated improvement is a 10–30x execution throughput increase. Combined with the gas limit increase from 60 million to 200 million per block, the theoretical capacity reaches approximately 10,000 TPS.
BALs also enable two secondary capabilities:
EIP-8159 provides the networking layer companion, enabling nodes to share access lists across the peer-to-peer network via the eth/71 protocol upgrade.
The parallelism approach differs from competitors. Solana's Sealevel uses native parallel scheduling. Monad employs optimistic execution with conflict detection. Ethereum's model requires explicit pre-declaration of dependencies — more conservative, but compatible with existing EVM tooling and the broader rollup ecosystem.
Glamsterdam bundles multiple gas repricing changes that collectively target a 78.6% reduction in average gas costs. The mechanics are distributed across several EIPs:
EIP-7904 performs a general repricing that realigns gas costs with actual computational resource consumption. Many current gas costs date to 2016-era hardware assumptions.
EIP-2780 reduces the intrinsic gas cost of standard ETH transfers, making simple sends up to 71% cheaper. A surcharge is added for transactions that create new accounts to offset state growth.
EIP-8037 introduces a reservoir model that separates computational gas from storage gas accounting. The target is to cap state growth at approximately 100 GiB per year at a 100 million gas limit, tying fees to actual data size created rather than a flat rate.
EIP-8038 increases gas constants for state-access opcodes that are currently underpriced, including EXTCODESIZE and EXTCODECOPY. These adjustments are necessary to prevent the repriced fee schedule from being exploited through cheap state reads.
The net effect: routine transactions become substantially cheaper, while operations that impose long-term costs on the network (state creation, cold storage reads) become more expensive. This aligns with the economic principle of pricing externalities — users who grow the state pay for the perpetual storage burden they impose.
As of the Ethereum Foundation's Checkpoint #9 report (April 10, 2026), development had progressed through the following stages:
Prior to the generalized devnet, testing was split across separate networks — one for ePBS (epbs-devnet) and one for BALs (bals-devnet). The merge into a unified environment represents the first time all components coexist and interact.
The path from generalized devnet to mainnet follows Ethereum's established sequence: devnet stabilization → client releases → dual security audits → public testnet deployment (Holesky, Sepolia) → mainnet fork announcement.
The Checkpoint #9 report characterized ePBS as "proving to be trickier than anticipated," noting that the change "touches practically everything" in the protocol stack. Gas repricing carries its own complexity, tracked through a dedicated technical todo list maintained by the development team.
No mainnet date has been committed. Developer documentation references June 2026 as aspirational. Independent assessments suggest Q3 or Q4 2026 is more likely.
Glamsterdam's impact extends beyond L1. Layer 2 rollups that settle on Ethereum benefit from two changes:
Daily L2 transactions have grown from 1 million in 2023 to 11 million in 2026, with 89% of all Ethereum ecosystem transactions now running through Layer 2 networks, according to L2Beat data. Cheaper L1 settlement strengthens Ethereum's position as the preferred settlement layer for rollups.
The deterministic factory predeploy (EIP-7997) at address 0x12 enables identical smart contract deployment across all EVM-compatible chains, reducing fragmentation for developers building cross-chain applications.
EIP-7708's mandatory logging of ETH transfers and burns improves on-chain accounting infrastructure — a prerequisite for institutional adoption, where audit trails and reconciliation are non-negotiable.
Implementation delay. Ethereum hard forks historically ship months behind initial targets. Pectra was delayed multiple times before its May 2025 activation. Glamsterdam's complexity — particularly the ePBS two-party coordination model — increases the probability of slippage.
Gas repricing breakage. Applications with hardcoded gas estimates will break under the new fee schedule. The transition requires ecosystem-wide tooling updates, and the adjustment period could cause temporary disruption.
BAL adoption. Parallel execution benefits depend on wallet and dApp tooling generating proper access lists. Until adoption reaches critical mass, the theoretical throughput gains may not materialize in practice.
Builder centralization post-ePBS. Removing relays eliminates one centralization vector but does not address builder-level concentration. Without FOCIL (deferred to Hegotá), sophisticated builders may still dominate block construction.
Competitive pressure. Solana's Alpenglow upgrade targets 150ms finality. Monad launched with 10,000+ TPS claims. If Glamsterdam slips to Q4 2026 or later, Ethereum's L1 competitiveness narrative weakens further, though its L2 ecosystem and security budget remain differentiators.
Glamsterdam represents the largest structural change to Ethereum since the Merge. The upgrade addresses two long-standing criticisms: L1 throughput limitations and off-chain infrastructure dependencies that concentrate power among a small number of relay operators and block builders.
The economic logic is sound. Enshrining proposer-builder separation removes a trust assumption that currently spans 80–90% of block production. Parallel execution via access lists unlocks hardware resources that sequential processing leaves idle. Gas repricing aligns fee schedules with 2026 hardware realities.
Execution risk remains the primary concern. ePBS touches every layer of the protocol stack, and the Ethereum Foundation's own assessment acknowledges the complexity. The timeline is aspirational, not committed. Whether Glamsterdam ships in Q2, Q3, or Q4 2026 will depend on devnet stability in the coming weeks.
The upgrade does not resolve all centralization concerns. Builder-level concentration persists without FOCIL, and the 30% OFAC compliance rate at the relay level highlights ongoing censorship resistance challenges. These issues are deferred, not dismissed — Hegotá is the next venue for addressing them.
For the broader Ethereum economy, Glamsterdam's value proposition is measurable: lower fees for users, lower settlement costs for L2s, and a reduction in off-chain trust assumptions. Whether the market prices that in before or after mainnet activation depends on devnet progress over the next 60–90 days.