Ethereum's Glamsterdam hard fork — the largest protocol upgrade since the September 2022 Merge — has entered its final devnet cycle. Glamsterdam-Devnet-7 launched July 14, 2026, and Devnet-8, expected in early August, will likely be the last internal testnet before public deployments on Sepolia (...
"Developers are now running devnets that include all planned EIPs... work is in its final phase before code hardening and public testnet deployment." — Parithosh Jayanthi, Ethereum Foundation Core Developer
Ethereum's Glamsterdam hard fork — the largest protocol upgrade since the September 2022 Merge — has entered its final devnet cycle. Glamsterdam-Devnet-7 launched July 14, 2026, and Devnet-8, expected in early August, will likely be the last internal testnet before public deployments on Sepolia (~August 25) and Hoodi (~September 8). Mainnet activation is tentatively targeted for October 8, 2026, though this date remains contingent on cross-client stability.
The upgrade bundles two headline EIPs: EIP-7732, which enshrines Proposer-Builder Separation (ePBS) directly into the consensus layer, and EIP-7928, which introduces Block-Level Access Lists (BALs) on the execution side. Together, they target a 3.3x gas limit increase from 60 million to 200 million per block, an estimated 78.6% reduction in Layer 1 fees through gas repricing, and a theoretical throughput ceiling of 10,000 transactions per second. Currently, 80–90% of Ethereum blocks are built through external relay infrastructure like MEV-Boost; ePBS moves that process on-chain.
The upgrade arrives at a moment of compressed staking yields. With 1.1 million active validators and roughly 34–41 million ETH staked (29–34% of supply), base APR sits at 2.78%. ETH trades near $1,832. The economic question for the network is whether tripling block capacity and compressing fees will generate enough incremental transaction volume to offset per-unit revenue declines — a familiar challenge from the post-Dencun period.
Glamsterdam, named after the combination of "Gloas" (consensus layer) and "Amsterdam" (execution layer) as specified in EIP-7773, represents the first Ethereum hard fork to simultaneously restructure both block construction and transaction execution.
The upgrade contains approximately ten EIPs spanning three categories:
Consensus Layer:
Execution Layer:
Gas Repricing Cluster:
Two late additions — EIPs 7688 and 8282 — remain under active development and may be finalized for Devnet-8.
Ethereum's current block production pipeline depends on an extra-protocol system: MEV-Boost, a middleware layer managed by Flashbots and other relay operators. According to network data, 80–90% of Ethereum blocks pass through these external relays. The arrangement works but introduces trust assumptions, censorship vectors, and latency dependencies that sit outside the protocol's formal security model.
EIP-7732 moves this separation on-chain. Under ePBS, the protocol formally distinguishes between the proposer (the validator who signs the consensus block) and the builder (the entity who assembles the execution payload and bids for inclusion). The trustless exchange of a block payload for payment is embedded directly into the protocol, removing the need for third-party relay infrastructure.
A new mechanism, the Payload Timeliness Committee (PTC), is introduced. A subset of validators attests whether the committed payload was revealed correctly and on time. This replaces the implicit trust in relay operators with an explicit, protocol-enforced verification step.
The projected impact: up to 70% reduction in MEV extraction leakage, according to estimates cited across multiple developer resources. Smaller, home-staked nodes gain the ability to compete more fairly with professional builders, reinforcing censorship resistance.
However, the PTC duty increases sensitivity to network latency. Validators operating nodes with poor connectivity may face attestation penalties. According to Benjamin Thalman, Staff Protocol Analyst at Figment, the upgrade "addresses two areas of particular interest to institutional stakers: the architecture of block building, and the speed at which validators can exit large staked ETH positions."
Block-Level Access Lists (BALs) represent the execution-layer counterpart to ePBS. The core problem: transaction execution cannot be parallelized without knowing in advance which addresses and storage slots each transaction will access. While EIP-2930 (Berlin, April 2021) introduced optional transaction-level access lists, they are not enforced and see limited adoption.
EIP-7928 introduces a mandatory block-level field — block_access_list_hash — containing the Keccak-256 hash of the RLP-encoded list of all accounts and storage locations accessed during block execution, along with their post-execution values.
This enables four forms of parallelization:
Historical analysis of Ethereum mainnet data shows that 60–80% of transactions access disjoint storage slots, making them immediately parallelizable. The remaining 20–40% can still benefit from post-transaction state diffs. Average BAL size in testing: approximately 70 KiB per block.
BALs are the prerequisite for safely raising the gas limit. Without parallel validation, a 200 million gas limit would overwhelm sequential block processing. With BALs, client teams can push block capacity toward that ceiling while maintaining acceptable validation times.
EIP-7904 recalibrates gas costs to reflect actual computational resource consumption on modern hardware. Many current opcode prices were set years ago and no longer correspond to real execution costs.
The recalibration follows a principle: computation-heavy operations get cheaper; state-access operations get more expensive.
Under EIP-8037's cost_per_state_byte model, parameterized in the bal-devnet-6 test network:
The net effect across the repricing cluster: an estimated 78.6% reduction in fees for both simple ETH transfers and complex smart contract interactions, according to developer benchmarks. For context, DeFi swaps currently cost $4–8 on Ethereum L1. A 78% reduction would bring those to roughly $0.88–$1.76.
The tension here is well-documented in Ethereum's post-Dencun experience. When EIP-4844 cut L2 data-posting costs by 90–99% in March 2024, blob fee revenue cratered. Total protocol revenue declined even as transaction counts rose. Glamsterdam's fee compression, if it follows a similar pattern, could further pressure validator economics unless accompanied by proportional volume increases.
The development trajectory, according to Christine Kim's ACDE #241 notes and multiple developer sources:
| Milestone | Date | Status | |-----------|------|--------| | Soldøgn Interop Devnet | May 2, 2026 | Completed | | Multi-client devnet (stable) | May 2026 | Completed | | Glamsterdam-Devnet-7 | July 14, 2026 | Completed; held under load | | Glamsterdam-Devnet-8 | Early August 2026 | Expected; includes EIP-8070 | | Sepolia public testnet | ~August 25, 2026 | Tentative | | Hoodi public testnet | ~September 8, 2026 | Tentative | | Mainnet activation | ~October 8, 2026 | Aspirational |
Devnet-8 will likely be the last internal devnet before public testnet upgrades. No separate short-lived testnet (analogous to "Mekong" from earlier upgrade cycles) is planned. Application developers are advised to begin testing on Devnet-7 immediately.
Historical precedent suggests caution on the timeline. Past forks have required two to four months of public-testnet seasoning. On that cadence, mainnet could land anywhere between October and December 2026.
Testing on Devnet-7 has included gas limits at 300 million — higher than the 200 million design target — to stress-test parallel execution under extreme conditions. This is described as "the highest increase in block size the network has ever seen."
For the network's 1.1 million active validators, Glamsterdam introduces mandatory client updates on both the consensus and execution layers. Nodes that fail to upgrade before mainnet activation will fork off the canonical chain.
Key operational changes:
PTC Duty: Validators join the Payload Timeliness Committee on a rotating basis, adding a new attestation obligation with latency sensitivity.
Faster Exits (EIP-8061): Exit and consolidation churn limits increase. Under current conditions, large staked positions can take weeks to clear the exit queue. Post-Glamsterdam, the same volume could process in days. The validator entry queue currently holds 2.5 million ETH with a 44-day waiting period.
Slashing Rules (EIP-8045): Slashed validators are excluded from block proposal duties, tightening protocol hygiene.
Hardware Considerations: Parallel disk reads may create a slight uptick in I/O requirements. However, the ePBS architecture ensures that block building complexity is absorbed by specialized builders, not proposers. This is designed to keep home-staking viable.
Glamsterdam's effects on L2 economics are indirect but consequential. The ePBS payload-propagation window creates room for higher blob capacity in subsequent Blob Parameter Only (BPO) forks. PeerDAS, activated with the Fusaka upgrade in December 2025, already enabled blob counts to begin scaling toward 48 per block. Glamsterdam extends this runway.
The gas repricing also affects L2 data-posting costs on L1. Rollups that post calldata (as opposed to blobs) will see cost changes from the opcode recalibration. Rollups already using blob transactions — the majority of major L2s — will primarily benefit from the broader capacity increases.
Since EIP-4844 in March 2024, L2 transaction costs dropped from $0.50–$5.00 to $0.001–$0.05. Glamsterdam does not directly alter blob pricing but creates the structural conditions for further capacity expansion.
Several features originally considered for Glamsterdam were moved to the subsequent Hegotá fork, targeted for late 2026:
The Ethereum Foundation also announced leadership transitions: Francesco D'Amato (researcher) is transitioning to EthLabs, while new protocol cluster leads include Will Corcoran (zkVM proofs, post-quantum consensus) and Kev Wedderburn (zkEVM development). Barnabé Monnot and Tim Beiko are in gradual management transitions. Alex Stokes is entering a leave period.
Glamsterdam represents Ethereum's most ambitious architectural change in four years. The simultaneous restructuring of block production (ePBS) and transaction execution (BALs) addresses two longstanding protocol constraints — centralized block building and sequential processing — in a single fork.
The engineering trajectory is on track. Multi-client devnets are stable, stress testing at 300 million gas limits exceeds the 200 million design target, and the public testnet schedule is tightening. The technical risk is concentrated in the final cross-client integration phase and the ePBS coordination complexity that developers have described as "slow but steady."
The economic implications are less certain. Ethereum's fee revenue has already compressed significantly since Dencun. A further 78% reduction in per-transaction costs requires a substantial increase in L1 transaction volume to maintain current revenue levels. Whether the combination of cheaper execution and parallel processing generates that volume — or simply accelerates value migration to L2s — remains the open question.
For validators, the upgrade is operationally demanding: dual-client updates, a new attestation role, and potential I/O changes. For institutional stakers, faster exit processing and fairer block production represent tangible improvements. For the broader ecosystem, the central bet is that Ethereum L1 can reclaim economic activity from both competing L1s and its own L2s by becoming meaningfully cheaper and faster — without sacrificing the decentralization properties that differentiate it.