Ethereum's Glamsterdam hard fork — the network's most significant protocol change since the September 2022 Merge — has entered public testnet deployment. The Sepolia testnet fork activated on August 3, 2026, with Hoodi scheduled for August 17 and mainnet targeted for September 16. Three architect...
"In Glamsterdam, Ethereum is getting ePBS, which lets proposers outsource to a free permissionless market of block builders." — Vitalik Buterin, Ethereum Co-Founder (March 2, 2026)
Ethereum's Glamsterdam hard fork — the network's most significant protocol change since the September 2022 Merge — has entered public testnet deployment. The Sepolia testnet fork activated on August 3, 2026, with Hoodi scheduled for August 17 and mainnet targeted for September 16. Three architectural changes anchor the upgrade: Enshrined Proposer-Builder Separation (EIP-7732), Block-Level Access Lists (EIP-7928), and a 10-EIP gas repricing package (EIP-8007). Together, they target a 78.6% reduction in L1 fees, a path to 200 million gas per block (up from the current 60 million), and up to 70% less MEV extraction through on-chain builder markets.
The stakes are structural. Over 88% of Ethereum blocks are currently built through off-chain MEV-Boost relays — a system that works but introduces trust assumptions and centralization risk outside protocol governance. Glamsterdam moves that process on-chain. Simultaneously, Block-Level Access Lists lay the technical groundwork for parallel transaction execution, the prerequisite for tripling the gas limit without proportionally increasing hardware demands on validators. The upgrade arrives as Ethereum L1 processes roughly 15 transactions per second at a 60 million gas limit; post-Glamsterdam, the theoretical ceiling rises to approximately 10,000 TPS — though real-world throughput depends on client optimization, validator hardware, and actual parallelism achievable in production blocks.
Glamsterdam is a combined consensus-layer ("Gloas") and execution-layer ("Amsterdam") hard fork, the third major upgrade in a 14-month sequence following Pectra (early 2025) and Fusaka (December 3, 2025). Where Pectra focused on user experience and validator consolidation, and Fusaka expanded data availability for rollups via PeerDAS, Glamsterdam targets L1 throughput and block-building architecture.
The upgrade carries approximately 15 EIPs across its two layers. Two headline proposals — EIP-7732 and EIP-7928 — deliver the structural changes. A third package of gas repricing EIPs under EIP-8007 recalibrates compute, memory, and storage costs. Additional EIPs address validator operations: EIP-8061 increases exit and consolidation churn for validators managing large staked positions, EIP-8045 excludes slashed validators from block proposals, and EIP-7688 introduces forward-compatible consensus data structures.
Notably absent: EIP-7782 (6-second slot times), deferred due to ZK-proving implications, and FOCIL (Fork-Choice Inclusion Lists, EIP-7805), pushed to the subsequent Hegotá upgrade to reduce implementation complexity.
The current Ethereum block production pipeline depends heavily on MEV-Boost, a sidecar protocol maintained by Flashbots. Over 88% of Ethereum blocks pass through this off-chain relay infrastructure. Proposers (validators) outsource block construction to specialized builders who bid for the right to assemble transaction bundles — a process that functions but sits entirely outside protocol rules.
EIP-7732 enshrines this proposer-builder separation directly into the consensus layer. Under the new model:
The data propagation window expands from approximately 2 seconds to roughly 9 seconds as a consequence of this restructuring. That additional time is not wasted — it allows the network to safely process larger blocks with more transactions.
The MEV implications are significant. According to multiple analyses cited in upgrade documentation, ePBS reduces MEV extraction by up to 70% by moving competition on-chain, making it transparent and auditable rather than routed through private off-chain coordination channels. MEV-Boost relays become optional rather than required. The upgrade does not eliminate MEV — it restructures the market around it.
For solo validators and home stakers, ePBS removes the competitive disadvantage of not running specialized relay infrastructure. Block proposal becomes more equitable at the protocol level.
Ethereum's current execution model is sequential: transactions execute one after another, each reading from and writing to a shared state tree. This is safe but slow. EIP-7928 introduces Block-Level Access Lists (BALs) — a record included in every block that maps every account and storage slot the block touches, along with post-execution state values.
The immediate benefit is enabling parallel transaction execution. Historical analysis of Ethereum mainnet data shows that 60–80% of transactions within a block access disjoint storage slots — meaning they do not conflict and can, in principle, execute simultaneously. The remaining 20–40% can be parallelized through post-transaction state diffs provided by BALs.
BALs also enable executionless sync, a new synchronization mode where nodes can verify blocks without replaying every transaction, dramatically reducing sync times for new nodes joining the network.
The gas limit increase — from 60 million currently to a target of 200 million — is not part of Glamsterdam itself. The upgrade provides the technical substrate. Once BALs are live and client implementations demonstrate stable parallel execution, the community can pursue gas limit increases through governance. At 200 million gas with effective parallelization, estimated throughput reaches approximately 10,000 TPS, compared to approximately 15 TPS today. These are projections, not guaranteed outcomes — actual throughput will depend on client performance, validator hardware capabilities, and real-world transaction patterns.
Under EIP-8007, Glamsterdam bundles 10 gas repricing proposals that collectively recalibrate what operations cost on Ethereum L1. The design philosophy: make computation cheaper while making permanent state creation more expensive. This targets state bloat — the growing size of Ethereum's state tree — while reducing fees for standard operations.
Key changes include:
The aggregate projected fee reduction is approximately 78.6% for typical transaction types, according to developer estimates in the EIP specifications. Actual fees will vary depending on L1 demand and the specific operations performed.
Glamsterdam introduces meaningful changes for Ethereum's validator set. All node operators must update both consensus-layer clients (Lighthouse, Prysm, Teku, Nimbus, Lodestar, Grandine) and execution-layer clients (Geth, Nethermind, Besu, Erigon, Reth) before mainnet activation. Precedent from Fusaka — where a Prysm client bug resulted in 382 ETH in lost rewards — underscores the operational risk of delayed updates.
The new Payload Timeliness Committee duty (from EIP-7732) adds a validator responsibility: PTC members attest to whether builders delivered their execution payloads on time. This requires reliable, low-latency infrastructure. Missed attestations become more costly as blocks carry higher value under the expanded gas capacity.
EIP-8061 increases the exit and consolidation churn rate, directly benefiting institutional stakers managing large positions. According to Figment's August 5, 2026 analysis, the upgrade will "materially improve exit liquidity" for validators — a critical factor for institutions that need predictable position sizing and withdrawal timelines.
For the parallel execution model, validators need to audit IOPS (input/output operations per second) capacity on their storage infrastructure. BALs increase the amount of state data referenced per block, and the eventual path to 200 million gas will place higher demands on disk throughput.
Glamsterdam creates a dual dynamic for Ethereum's Layer 2 ecosystem. L2 networks — Arbitrum, Optimism, Base, zkSync — originally attracted users because L1 gas was too expensive and MEV on L1 punished retail traders. If Glamsterdam delivers a 78% fee cut and up to 70% MEV reduction, some activity may migrate back to Ethereum's base layer, which carries stronger security guarantees than any rollup.
Current L2 fees are already low following Fusaka's PeerDAS upgrade: Base averages approximately $0.05 per transaction, zkSync Era roughly $0.07, and Arbitrum One and OP Mainnet approximately $0.09. The competitive gap narrows further if L1 fees drop by 78%.
However, L2s also benefit from Glamsterdam. They post data blobs to L1, and expanded block space reduces their settlement costs by an estimated 40%. The net effect depends on where the fee reduction attracts more marginal users — L1 at lower costs, or L2 at even-lower costs.
The economic question is whether Ethereum L1 recaptures enough fee revenue from increased throughput to offset lower per-transaction pricing. The precedent from EIP-4844 (Dencun, March 2024) is instructive: L2 data costs dropped approximately 90%, but L1 fee revenue declined because the freed capacity was consumed at lower price points. Glamsterdam's parallel execution and ePBS are more structurally different — they increase the volume ceiling rather than just reducing data costs — but the revenue outcome remains uncertain.
| Milestone | Date | |---|---| | Final devnet stage reached | Mid-June 2026 | | Sepolia testnet fork | August 3, 2026 | | Hoodi testnet fork | August 17, 2026 | | Mainnet activation target | September 16, 2026 | | Fallback window | Q4 2026 |
Three risks could push activation beyond September:
The meta specification (EIP-7773) remains in Draft status. Ethereum Foundation contributors have noted that Glamsterdam is "proving trickier and slower than Fusaka."
Glamsterdam represents Ethereum's attempt to reclaim L1 relevance after two years of ceding execution to rollups. The upgrade does not increase the gas limit directly — it provides the architecture (ePBS, BALs, repricing) that makes a future increase to 200 million gas feasible. Whether Ethereum L1 recaptures meaningful transaction volume depends on execution: client stability across 11 implementations, validator infrastructure readiness, and whether the fee reduction proves sufficient to compete with sub-$0.10 L2 transactions. The Sepolia testnet fork on August 3 was the first live test. Hoodi on August 17 and mainnet in September will determine whether Ethereum's most ambitious post-Merge upgrade delivers or slips.