Ethereum's Glamsterdam hard fork — the network's largest protocol change since the 2022 Merge — is in final devnet testing ahead of a targeted mid-2026 mainnet activation. The upgrade bundles eight Ethereum Improvement Proposals (EIPs) anchored by two structural changes: EIP-7732, which enshrines...
"The Glamsterdam upgrade is a complete structural overhaul aimed at solving two of Ethereum's oldest problems: centralization and state bloat." — Vitalik Buterin, Ethereum Co-Founder
Ethereum's Glamsterdam hard fork — the network's largest protocol change since the 2022 Merge — is in final devnet testing ahead of a targeted mid-2026 mainnet activation. The upgrade bundles eight Ethereum Improvement Proposals (EIPs) anchored by two structural changes: EIP-7732, which enshrines proposer-builder separation (ePBS) directly into the consensus layer, and EIP-7928, which introduces block-level access lists enabling parallel transaction execution on the EVM for the first time.
Core developers reached consensus on a post-upgrade gas limit floor of 200 million per block, up from the current 60 million — a 3.3x expansion. Combined with parallel execution, the upgrade targets approximately 10,000 transactions per second on Layer 1, roughly 10x current throughput. Gas fee reductions of approximately 78% are projected across both simple transfers and complex smart contract interactions, according to devnet benchmarks. As of late May 2026, the Ethereum Foundation confirmed devnets are running stably with ePBS functioning across multiple client implementations. Public testnet activations on Holesky and Sepolia remain the final gate before a mainnet date is set.
The upgrade arrives at a critical juncture. Ethereum processes approximately 1.7 million daily L1 transactions at an average fee of $0.21, while its 73 active Layer 2 rollups collectively handle roughly 5,600 TPS with $48 billion in total value locked. The network's economic model faces pressure: L2s contributed just $4.9 million in blob fees to mainnet despite earning $94 million in profit on Base alone over a 180-day period. Glamsterdam's capacity expansion aims to make L1 competitive enough to recapture fee revenue while simultaneously lowering costs for L2 data posting.
Ethereum's block production currently depends on MEV-Boost, a third-party relay system operated by a handful of entities. According to on-chain data, approximately 93% of Ethereum blocks use MEV-Boost, with just four relay operators controlling effectively all relay traffic as of 2024. This creates a single point of failure and a centralization vector that runs counter to the network's design philosophy.
EIP-7732 moves the proposer-builder handoff directly into the Ethereum protocol. Under the current system, a validator (the proposer) outsources block construction to a builder via an external relay, trusting the relay to mediate the exchange honestly. Under ePBS, this exchange is trustless and protocol-enforced. The proposer commits to a block header, and the builder reveals the execution payload — all verified by the consensus layer without intermediary software.
The structural benefit extends beyond decentralization. By separating the execution payload from the consensus block, ePBS expands the data propagation window from approximately 2 seconds to roughly 9 seconds. This is the mechanism that unlocks the network's ability to safely handle higher gas limits and more data blobs per block for Layer 2 networks.
The Ethereum Virtual Machine currently processes transactions sequentially — one after another, in order. EIP-7928 introduces block-level access lists that pre-declare which accounts and smart contracts each transaction in a block will interact with. This metadata allows the EVM to identify non-conflicting transactions and execute them simultaneously across multiple processing threads.
The result, according to devnet testing, is a 3-4x throughput improvement without increasing block size. Transactions that touch different state can run in parallel, while those with dependencies remain serialized. This is the execution-layer complement to ePBS: one expands the available time window, the other fills it with more computation.
The consensus among core contributors on a 200 million gas limit floor post-Glamsterdam represents the largest single capacity expansion in Ethereum's history. The current 60 million limit has been the practical ceiling since validators voluntarily raised it in 2024.
However, a 3.3x gas limit increase without guardrails would accelerate state growth — the accumulation of contract storage and account data that every full node must maintain. This is where EIP-8037 provides the counterbalance.
EIP-8037 raises the cost of writing new state to the chain. Contract deployment costs increase approximately 10x, and new account creation rises roughly 8.5x. Core teams agreed to drop a proposed dynamic pricing mechanism in favor of a fixed cost_per_state_byte model, with future repricing handled at fork boundaries rather than within a fork.
The target: state growth capped at approximately 60 GiB per year at a 300 million gas block limit. This ensures the higher gas limit translates into more computation throughput, not unbounded storage expansion. The tradeoff is explicit — applications that create large amounts of new state will pay significantly more, while applications that primarily read or modify existing state benefit from lower fees overall.
The upgrade has progressed through several development phases:
Remaining milestones before mainnet:
June 2026 remains the aspirational target. Most core developers now consider Q3 2026 (July-September) the more realistic window, according to post-Soldøgn developer commentary. The teams have explicitly stated that correctness takes priority over schedule.
Three significant features originally scoped for Glamsterdam have been deferred to the subsequent Hegotá fork, planned for late 2026: FOCIL (fork-choice inclusion lists), Verkle Trees, and account-abstraction upgrades.
Glamsterdam's capacity expansion addresses a structural economic problem. Ethereum's fee revenue has declined significantly since the March 2024 Dencun upgrade introduced blob transactions for L2 data posting, which reduced L2 costs but also reduced ETH burn from fee destruction.
Current Ethereum economic snapshot (May 2026):
The 78% gas fee reduction projected from Glamsterdam could push L1 transaction costs to $0.05-0.18, making Ethereum L1 cost-competitive with some Layer 2 environments for the first time since 2020. This creates a direct competitive dynamic: L2 rollups that currently charge $0.001-0.05 per transaction retain a cost advantage, but the gap narrows substantially for complex DeFi operations where L1 composability carries a premium.
The risk is that lower fees, combined with higher throughput, may not generate enough aggregate revenue to offset the per-transaction decline — the same dynamic that played out post-Dencun. Industry estimates suggest blob fees could contribute 30-50% of total ETH burn by late 2026, but that volume has not materialized at the scale needed to restore pre-Dencun burn rates.
Ethereum currently operates approximately 1.1 million active validators. Glamsterdam changes the economics of block production for these operators in several ways.
MEV Revenue: Protocol-level PBS integration is projected to reduce centralized MEV extraction by approximately 70%, according to Figment's institutional staking analysis. For validators, this means MEV earnings become more stable and predictable, though the near-term impact depends on adoption rates of the new ePBS market. The existing MEV-Boost relay system continues to function — ePBS is strictly opt-in, with no forced migration.
Block Selection Authority: Under ePBS, validators retain full authority over block selection without needing to trust external relay operators. This is particularly relevant for institutional stakers subject to compliance requirements around transaction ordering.
Execution Layer Rewards: A more competitive, trust-minimized builder market could be positive for the variable execution layer component of staking rewards, as increased builder competition generally improves block optimization over time.
Liquid Staking Concentration: The current staking landscape remains consolidated. Lido controls approximately 28% of staked ETH, Coinbase cbETH holds 14%, and EigenLayer's restaking captures 11%. Glamsterdam does not directly address this concentration, though the removal of relay dependencies may lower barriers for smaller staking operations.
Glamsterdam does not exist in a vacuum. Solana's Alpenglow consensus overhaul, currently in validator testing as of late May 2026, targets sub-second latency reductions. The two largest blockchains are simultaneously undertaking their most significant architectural changes since launch.
L2 ecosystem metrics (May 2026):
If Glamsterdam delivers its 10,000 TPS target on L1, Ethereum mainnet alone would process nearly double the current combined L2 throughput. This raises a strategic question the Ethereum community has debated since the rollup-centric roadmap was adopted: whether a sufficiently fast L1 reduces the value proposition of general-purpose L2s, or whether the two layers serve structurally different markets.
The economic-value perspective suggests the answer is nuanced. L2s that serve as application-specific environments with customized execution (e.g., gaming, social, institutional) retain differentiated value. General-purpose L2s competing primarily on cost face margin compression if L1 fees fall below $0.10.
Glamsterdam represents Ethereum's most significant architectural change since the Merge, addressing two structural limitations — sequential execution and relay-dependent block building — that have constrained the network for years. The technical scope is ambitious: eight EIPs, a 3.3x gas limit expansion, and the first-ever protocol-level separation of block proposing and building.
The development trajectory is measured. The Soldøgn Interop demonstrated multi-client stability, but public testnet activations and dual audit phases remain before a mainnet date can be set. The deferral of Verkle Trees and FOCIL to Hegotá reflects a deliberate scoping decision — ship the capacity expansion now, handle state structure and censorship resistance in a subsequent fork.
The economic question is whether higher throughput at lower per-transaction cost generates sufficient aggregate fee revenue to sustain the network's security budget. This is not a new question — it emerged after Dencun and intensified as L2s grew to handle the majority of Ethereum ecosystem transactions while contributing a fraction of fee revenue back to L1. Glamsterdam does not resolve this tension. It expands the pie and reduces the price per slice, betting that volume growth will compensate.
For the 1.1 million validators, 73 active L2 rollups, and $44.8 billion in liquid-staked ETH that constitute the Ethereum economy, the upgrade is the most consequential protocol event since January 2024's Dencun. The market will price the outcome when Holesky goes live — the first tangible signal of a mainnet date.