Ethereum's Glamsterdam upgrade — combining the Amsterdam execution layer and Gloas consensus layer forks — activated on the Sepolia testnet October 6, 2026, and is scheduled for Hoodi testnet activation at epoch 132,352 on October 26, 2026 at 17:42:48 UTC. No mainnet date has been set; early Dece...
"I can just spin up a thousand builders, rotate them, offer very high bids, and not produce payloads." — Potuz, Ethereum Consensus Developer
Ethereum's Glamsterdam upgrade — combining the Amsterdam execution layer and Gloas consensus layer forks — activated on the Sepolia testnet October 6, 2026, and is scheduled for Hoodi testnet activation at epoch 132,352 on October 26, 2026 at 17:42:48 UTC. No mainnet date has been set; early December 2026 remains the working target.
The upgrade contains two headline proposals: EIP-7732, which enshrines proposer-builder separation (ePBS) directly into the consensus protocol, and EIP-7928, which introduces block-level access lists (BALs) enabling parallel state reads. Together they represent the most significant structural change to Ethereum's block production pipeline since the Merge in September 2022. A concurrent gas limit raise from 60 million to 200 million on testnets — more than 3x current mainnet capacity — is being stress-tested alongside the fork.
The stakes are material. MEV-Boost relays currently mediate approximately 92% of Ethereum's staked validator block production. Three builders — Titan (52.16%), BuilderNet (24.63%), and Quasar (15.06%) — assemble roughly 92% of MEV-Boost blocks. EIP-7732 aims to replace the trusted relay layer entirely with an in-protocol auction mechanism. Academic research has identified a "free option problem" that could cause builders to strategically withhold payloads on 0.82% of blocks under normal conditions, rising to 6% during high-volatility periods.
Glamsterdam bundles at least 19 EIPs as of the August 16 scope freeze, according to EIP-7773. The full list spans execution, consensus, and tooling layers. Core proposals include:
| EIP | Name | Layer | |-----|------|-------| | EIP-7732 | Enshrined Proposer-Builder Separation (ePBS) | Consensus | | EIP-7928 | Block-Level Access Lists (BALs) | Execution | | EIP-2780 | Resource-based intrinsic transaction gas | Execution | | EIP-7688 | Forward-compatible consensus data structures | Consensus | | EIP-7708 | ETH transfers emit a log | Execution | | EIP-7778 | Block gas accounting without refunds | Execution | | EIP-7843 | SLOTNUM opcode | Execution | | EIP-8037 | Increased and separately metered state creation cost | Execution | | EIP-8038 | Updated state-access costs | Execution |
The naming convention follows Ethereum's dual-layer pattern: "Amsterdam" (execution, named after a city) merged with "Gloas" (consensus, named after a star) produces "Glamsterdam." This is the first upgrade since the Merge to fundamentally alter how blocks are proposed, built, and validated.
Since the Merge, Ethereum's block production has relied on an out-of-protocol system called MEV-Boost, developed by Flashbots. Under MEV-Boost, validators outsource block building to specialized builders through trusted relays. The system works, but it introduces centralization vectors and trust assumptions that sit outside protocol governance.
The numbers illustrate the concentration. As of October 2026, approximately 92% of Ethereum's staked validators use MEV-Boost. Relay market share is dominated by Ultra Sound Relay (35.1%), bloXroute regulated (23.8%), Titan Relay (24.19%), and Aestus (10.03%), according to Relayscan data. On the builder side, three entities — Titan, BuilderNet, and Quasar — assemble approximately 92% of all MEV-Boost blocks. Flashbots' own relay has shrunk to 2.4% market share.
EIP-7732 moves the proposer-builder handshake into the consensus protocol itself. Under ePBS:
The practical effect: relays become unnecessary as intermediaries. The protocol enforces the deal between proposer and builder directly. This removes a class of trusted middleware that currently handles billions of dollars in MEV extraction annually. MEV-Boost currently adds 10–30% to validator base rewards, according to industry estimates.
On the censorship resistance front, data from MEV Watch shows OFAC-filtering relays now account for 16.7% of MEV-Boost payloads as of October 14, 2026 — down 44% over the prior 30 days. Neutral, non-filtering relays handle roughly 70% of daily block production. ePBS, by removing the relay layer entirely, would further reduce the surface area for transaction filtering at the infrastructure level.
EIP-7928 introduces Block-Level Access Lists (BALs): a per-block record of every account and storage slot accessed during execution, along with post-execution state values.
This addresses a fundamental bottleneck. Ethereum's current execution model processes transactions sequentially. Each transaction must read and write to the global state trie in order, creating I/O contention and limiting throughput regardless of available hardware resources.
BALs change this by recording access patterns at the block level. Once a block's access list is known, nodes can:
The parallelization benefits compound with the gas limit increase. At 200 million gas per block — 3.3x the current 60 million mainnet limit — sequential execution would strain node hardware. BAL-driven parallel reads are designed to keep node requirements manageable even as block sizes grow.
Ethereum's mainnet currently operates at a 60 million gas limit per block, doubled from 30 million during the 2025 validator-driven increase. Glamsterdam's testnet deployments are simultaneously testing a further raise to 200 million gas.
On Sepolia, following the October 6 activation, blocks began producing near the 200 million gas ceiling. Early data showed utilization between 26% and 46% of capacity, indicating the network was not saturated under test conditions. A private rehearsal on Devnet-11 completed the transition from 60 million to 200 million without losing finality.
The 200 million figure is not a protocol mandate. It is a configurable validator preference. After mainnet activation, each validator can independently signal their preferred gas limit. The network converges on a target through a gradual adjustment mechanism where validators vote the limit up or down by 1/1024th per block.
If adopted on mainnet, a 200 million gas limit would represent a 233% increase in raw block capacity. For context: Ethereum's gas limit history shows three major steps — 15 million target/30 million max under EIP-1559 (August 2021), 30 million effective (2021–2024), and 60 million (2025–present). A move to 200 million would be the largest single capacity expansion in the network's history.
The Ethereum Foundation's Protocol Cluster published the Hoodi testnet announcement on October 16, 2026, confirming activation parameters. The rollout timeline:
| Milestone | Date | Status | |-----------|------|--------| | Devnet-11 rehearsal | Pre-October 2026 | Completed | | Sepolia activation (epoch 353,024) | October 6, 2026 | Completed | | Client releases for Hoodi | October 16, 2026 | Published | | Hoodi activation (epoch 132,352) | October 26, 2026 | Scheduled | | Mainnet activation | TBD (early December target) | Unconfirmed |
Client readiness for Hoodi spans both layers. Consensus clients with published releases: Grandine, Lighthouse, Lodestar 1.50.0, Nimbus 26.10.1, Prysm, and Teku 26.10.1. Execution clients: Besu 26.10.0, Ethrex 30.0.0-rc.2, Geth, and Nethermind 2.2.0. The Prysm team issued a patch during Sepolia testing to address issues encountered during the activation.
The All Core Developers Consensus call #189 on October 15 — a 90-minute session — covered Glamsterdam status, Hegotá planning, and improvements to the hard fork development process. Candidate mainnet dates of December 1, 8, and 9 have been floated, but no decision was finalized.
The most substantive technical concern around ePBS is the "free option problem," analyzed in a September 2025 academic paper (arXiv:2509.24849). Under ePBS, once a proposer commits to a builder's bid, the builder gains a short window in which they can choose not to deliver the execution payload. If they withhold, the slot produces an empty block — the proposer still earns the consensus reward, but no transactions are processed.
This creates what researchers describe as a free option: the builder observes market conditions after winning the auction and can decide whether delivering the payload is still profitable. If prices have moved against them, they simply do not deliver. The cost is an empty slot; the builder faces no additional penalty beyond forfeiting the bid.
The paper estimates this would affect approximately 0.82% of blocks under normal volatility conditions, rising to roughly 6% during high-volatility periods. On a network producing approximately 7,200 blocks per day, that translates to 59 empty slots daily in normal markets and 432 during volatile episodes.
During testnet preparation, consensus developer Potuz raised a related concern specific to testnets: on networks with free test ETH, an attacker could "spin up a thousand builders, rotate them, offer very high bids, and not produce payloads" — effectively griefing the auction mechanism at no cost. While this is primarily a testnet issue (mainnet builders have real capital at risk), it highlights the mechanism's sensitivity to builder behavior.
Proposed mitigations include builder bond requirements, reputation systems, and PTC-based penalties. None are included in Glamsterdam's current scope. The Ethereum community appears to be shipping ePBS first and iterating on anti-abuse mechanisms in subsequent upgrades.
Ethereum developers have already named the next upgrade: Hegotá, combining the "Bogotá" execution layer (named after the Devcon city) with the "Heze" consensus layer (named after a star). Following Glamsterdam's slip from H1 to Q4 2026, Hegotá is now expected in 2027.
The Hegotá scope is not finalized, but early discussions center on:
The Hegotá scope is expected to be discussed in detail following Glamsterdam's mainnet activation. Sixty-six proposals were under consideration as of the latest developer meeting summaries.
Glamsterdam rewrites two of Ethereum's most consequential subsystems — block building and state access — in a single fork. The economic implications are significant: ePBS removes an entire category of trusted intermediary from the MEV supply chain, while BALs lay the groundwork for parallel execution that could make 200 million gas blocks feasible on consumer hardware.
The upgrade has cleared Sepolia and is nine days from its Hoodi testnet fork. Client releases are published. The remaining question is mainnet timing and whether the free option problem proves manageable without explicit penalties.
For validators, node operators, and infrastructure providers, the preparation window is narrowing. Glamsterdam-compatible client versions are available now. The Ethereum Foundation has stated plainly: the mainnet activation time will be announced once decided by client teams.