Ethereum's Glamsterdam hard fork — the network's largest protocol change since the 2022 Merge — activates on the Sepolia testnet on October 6, 2026, at epoch 353,024, slot 11,296,768 (13:53:36 UTC). The upgrade bundles 18 Ethereum Improvement Proposals across two layers: the Amsterdam execution-l...
"It's really not just a blockchain anymore. It's a hybrid architecture that combines together blockchains and modern cryptography, to enable much more powerful properties." — Vitalik Buterin, Ethereum Co-Founder
Ethereum's Glamsterdam hard fork — the network's largest protocol change since the 2022 Merge — activates on the Sepolia testnet on October 6, 2026, at epoch 353,024, slot 11,296,768 (13:53:36 UTC). The upgrade bundles 18 Ethereum Improvement Proposals across two layers: the Amsterdam execution-layer package and the Gloas consensus-layer package. Its two headline features — enshrined proposer-builder separation (ePBS, EIP-7732) and block-level access lists (BALs, EIP-7928) — restructure how blocks are built, validated, and processed.
The economic stakes are material. Glamsterdam raises the block gas limit from 60 million to 200 million, targets a roughly 78.6% reduction in L1 gas fees, and lays the groundwork for parallel transaction execution that could eventually push L1 throughput toward 10,000 TPS — up from the current average of 25.9 TPS recorded in Q2 2026. For 1.24 million active validators securing approximately 39.7 million staked ETH, the upgrade rewires how MEV revenue flows by replacing off-chain relay infrastructure with protocol-native auctions. Mainnet deployment is expected in Q4 2026, though no official date has been set.
Glamsterdam merges two upgrade packages into a single hard fork. The Amsterdam execution-layer upgrade modifies transaction processing, gas accounting, and smart contract capabilities. The Gloas consensus-layer upgrade alters block validation, builder-proposer dynamics, and validator operations.
The 18 scheduled EIPs span several categories:
Block Construction & Validation:
Gas & Storage Pricing:
Execution Layer Enhancements:
Consensus Layer Changes:
Both execution and consensus layer clients require updates. Six consensus clients (Grandine 3.0.0-rc.0, Lighthouse 8.3.0-rc.0, Lodestar 1.49.0, Nimbus 26.10.0, Prysm 7.2.0, Teku 26.9.1) and six execution clients (Besu 26.9.0, Ethrex 29.0.1, Erigon 3.7.1, go-ethereum 1.17.7, Nethermind 2.1.0, Reth 2.7.0) have released or are releasing Sepolia-compatible versions.
The current Ethereum block production pipeline relies on MEV-Boost, an off-chain middleware system where validators outsource block building to specialized builders via third-party relays. As of September 2026, three to five block builders — Beaverbuild, Titan, Rsync, BuilderNet, and Flashbots Builder — win more than 85% of Ethereum L1 slots. Flashbots' own relay, once dominant, now carries only 2.4% of daily block production. Non-filtering, neutral relays handle approximately 70% of blocks.
EIP-7732 moves this entire mechanism into Ethereum's consensus protocol. Under ePBS:
The separation of consensus validation from execution validation gives validators additional time to verify execution payloads. According to multiple technical analyses, the block propagation window shrinks from nine seconds to approximately two seconds.
The economic consequence: MEV revenue that previously flowed through opaque off-chain auction systems now routes through transparent on-chain mechanisms. This does not eliminate MEV — the economic incentive to reorder, front-run, and sandwich transactions persists as long as profitable ordering opportunities exist. What changes is the settlement layer. Validators no longer need sophisticated block-building infrastructure; they only validate and select bids.
EIP-7928 introduces Block-Level Access Lists, structured declarations embedded in the block header that specify exactly which accounts and storage slots each transaction will read or write. A hash of this access list — the "BAL root" — is stored directly in the block header.
The technical mechanics:
Ethereum L1 currently averages 25.9 TPS, with a real-time measurement near 17.49 TPS and a theoretical maximum of approximately 238 TPS under the existing 60 million gas limit. Glamsterdam's 200 million gas limit, combined with BAL-enabled parallel processing, targets a roughly tenfold throughput increase. The Ethereum Foundation's "Strawmap" roadmap aims for 10,000 TPS by 2029 through a sequence of seven hard forks.
During Devnet-11 testing, Nethermind's execution client processed 570.7 billion gas in 3 minutes and 15 seconds, completing 2,302 tests under the new rules. These results are preliminary — testnet conditions differ from mainnet load — but indicate the infrastructure can handle substantially higher gas throughput.
Glamsterdam bundles a third round of gas repricing changes. EIP-8037 increases and separately meters state creation costs, while EIP-8038 updates state-access gas costs. Combined with the 200 million gas limit, the repricing is projected to reduce average L1 fees by approximately 78.6%.
These changes affect application developers directly. Contracts relying on fixed gas stipends or hardcoded gas limits may require modification. The Ethereum Foundation's testnet announcement specified that "application developers must test contracts against new gas rules."
The fee reduction has implications for Ethereum's economic model. Lower fees reduce the amount of ETH burned through EIP-1559's base fee mechanism. With 39.7 million ETH staked (roughly one-third of circulating supply) and base staking yields averaging 2.78%, reduced burn could affect ETH's supply dynamics — though the magnitude depends on how much additional transaction volume the lower fees attract.
As of October 2026, Ethereum's validator set comprises approximately 1.24 million active validators, up roughly 96,000 since January 2026. Close to 39.7 million ETH sits in the deposit contract. The entry queue shows 1,508,343 ETH with a 26-day wait; the exit queue holds 847,849 ETH with a 14-day, 17-hour wait.
Base staking APR sits at approximately 2.78%, compressed from the 4%+ yields available in 2023. MEV-Boost adds 0.5% to 1.0% for operators running relay software, bringing realistic all-in yield to 3.3% to 3.8%.
Glamsterdam alters this calculus in several ways:
Reduced operational burden. Validators no longer need to build blocks; they validate and select builder bids. This lowers hardware and bandwidth requirements, particularly for solo stakers who lack the infrastructure to compete with professional builders.
MEV transparency. On-chain auctions replace off-chain relay negotiations. Revenue that previously required MEV-Boost middleware now flows through protocol-native channels. Whether this increases or decreases total MEV revenue for validators depends on builder competition dynamics post-ePBS.
Consolidation effects. EIP-7251, implemented earlier, raised the maximum effective balance per validator from 32 ETH to 2,048 ETH. Large operators are merging many 32-ETH validators into fewer high-balance nodes, reducing active key counts while maintaining or increasing total staked ETH.
Ethereum developers have flagged a specific attack vector ahead of the Sepolia testnet launch. Under ePBS, a malicious actor can create disposable builder identities, submit artificially high bids to win block construction rights, and then withhold the promised execution payload.
Potuz, an Ethereum consensus developer, stated during discussions: "I can just spin up a thousand builders, rotate them, offer very high bids, and not produce payloads." He added: "Any teenager can do this."
The attack is particularly acute on testnets where test ETH has no economic cost. On mainnet, builders who withhold payloads forfeit their bids — a financial deterrent that does not exist in testnet environments. Current circuit breakers fall back to locally-built blocks only after several payload failures, but this defense is insufficient against attackers cycling through fresh identities.
Developer Fredrik Svantes referenced the standard upgrade protocol requiring "at least 14 days between release-ready client software and the first public testnet activation" for security reviews and bug-bounty exposure. Core developer Alex Stokes urged client teams to release software as early as possible to maximize reviewer scrutiny.
The Hoodi testnet activation is expected by October 27. How ePBS performs under adversarial conditions on Sepolia will determine whether the mainnet timeline holds.
Glamsterdam does not exist in isolation. Buterin wrote in a September 27, 2026, blog post that 2027's planned Hegotá fork is "likely to be Ethereum's last 'normal' fork" before the network transitions toward proof-heavy, cryptographic infrastructure — a shift Buterin characterizes as Ethereum evolving from a blockchain into a "hybrid architecture that combines together blockchains and modern cryptography."
The Ethereum Foundation's Strawmap outlines seven hard forks over 18 months following Glamsterdam, targeting 10,000 TPS by 2029. Glamsterdam's ePBS and BALs are foundational to this trajectory — ePBS restructures block production to support higher throughput, while BALs provide the data structures needed for parallel execution at scale.
For the broader L1 ecosystem, Glamsterdam's throughput and fee implications affect competitive dynamics. Solana processes approximately 4,000 TPS in production. A 200 million gas limit with parallel execution narrows that gap meaningfully, though Ethereum's actual throughput post-Glamsterdam will depend on real-world adoption of BAL-optimized block construction.
Glamsterdam restructures two foundational elements of Ethereum's architecture: how blocks are built and how transactions are processed. The shift from relay-dependent MEV auctions to protocol-native ePBS removes a centralization vector that has concentrated 85%+ of block production in fewer than five builders. BALs introduce the data structures required for parallel execution, though the full throughput gains will materialize across subsequent forks.
The economic value at stake is substantial. Approximately $95 billion in staked ETH secures the network. MEV revenue, staking yields, and transaction fee economics all shift under the new framework. Whether Glamsterdam delivers on its throughput and fee projections depends on testnet performance over the coming weeks — starting with Sepolia on October 6.
The data will speak on the testnet. The market will speak on mainnet.