Ethereum's Glamsterdam hard fork, now targeted for Q3 2026 after slipping from its original H1 window, represents the most significant structural overhaul of the network's execution and consensus layers since The Merge. The upgrade enshrines Proposer-Builder Separation (ePBS) into the protocol vi...
"No amount of talking about Ethereum's roadmap and vision matters if we cannot achieve coordination levels that consistently meet goals on schedule." — Tomasz K. Stańczak, Co-Executive Director, Ethereum Foundation
Ethereum's Glamsterdam hard fork, now targeted for Q3 2026 after slipping from its original H1 window, represents the most significant structural overhaul of the network's execution and consensus layers since The Merge. The upgrade enshrines Proposer-Builder Separation (ePBS) into the protocol via EIP-7732, replacing off-chain relay infrastructure that currently handles approximately 90% of Ethereum blocks. Simultaneously, Block-Level Access Lists (EIP-7928) lay the groundwork for parallel transaction execution, targeting a gas limit increase from 60 million to 200 million per block — a 3.3x expansion.
At current throughput of roughly 20 TPS on L1, Ethereum processes a fraction of competitor networks' capacity. Glamsterdam targets 10,000 TPS and a projected 78% reduction in L1 gas fees. The upgrade also restructures MEV economics: three builders currently produce over 80% of MEV-Boost blocks, a concentration that ePBS is designed to dilute by moving the builder auction on-chain. Gas repricing EIPs (EIP-8037, EIP-7976, EIP-7981) raise state-creation costs to prevent runaway state bloat under higher limits, while EIP-7778 removes gas refunds from block accounting to improve builder bid predictability.
The economic stakes are material. Ethereum L1 generates approximately $65 million in annualized fee revenue against $5–8 billion in total ecosystem value flows, according to webthreepedia's foundational economic analysis. Glamsterdam's capacity expansion could compress L1 fees further in the short term while potentially increasing aggregate throughput revenue. For institutional stakers managing 35.9 million staked ETH (28.9% of supply), the upgrade alters exit liquidity timelines, MEV reward distribution, and validator hardware requirements.
Glamsterdam combines consensus-layer changes ("Gloas") with execution-layer changes ("Amsterdam") across three primary vectors:
1. Enshrined Proposer-Builder Separation (EIP-7732): Moves block-builder coordination from external relays into the protocol itself, expanding the payload propagation window from 2 seconds to approximately 9 seconds.
2. Block-Level Access Lists (EIP-7928): Introduces per-block state maps recording every account and storage slot accessed during execution, enabling clients to process non-conflicting transactions concurrently across multiple CPU cores.
3. Gas Schedule Repricing (EIP-7904, EIP-8037, EIP-7976, EIP-7981, EIP-7778): A suite of cost adjustments that raise state-creation fees, increase the calldata floor, recalibrate access list pricing, and remove gas refunds from block accounting.
Additional EIPs include EIP-7708 (native ETH transfers emit ERC-20-format logs), EIP-7954 (raises the EIP-170 contract size limit), and EIP-8024 (backward-compatible stack opcodes). Two proposals were explicitly excluded: EIP-7782 (6-second slots), declined due to conflicts with ZK proving timelines, and FOCIL (Fork-Choice Enforced Inclusion Lists), deferred to the successor fork Hegotá.
According to Everstake's analysis, the upgrade was formally scoped following the Soldøgn interop event on May 2, 2026, where over 100 core developers reached consensus on the EIP set.
The current Ethereum block production pipeline depends on off-chain relays operated by a handful of entities. As of June 2026, relay market share stands at: Ultra Sound relay (31.4%), Titan Relay (24.0%), bloXroute Max-Profit (15.5%), and bloXroute Regulated (14.4%), according to Relayscan data. Flashbots' own relay handles just 3.4% of payloads.
On the builder side, concentration is more acute. Titan produces 55.2% of all MEV-Boost blocks, followed by Quasar (21.4%) and Eureka (11.1%). Three builders control roughly 87% of block production. This concentration creates censorship vectors: over 50% of high-value Ethereum transactions were routed through private channels as of mid-2025, according to Datawallet research.
EIP-7732 replaces this architecture with an in-protocol commit-reveal mechanism. Builders assemble blocks and cryptographically seal their contents, publishing bids with payload commitments. Proposers select the highest-paying bid without the ability to inspect or tamper with transaction contents. A new validator duty — the Payload Timeliness Committee (PTC) — attests to payload delivery, removing the need for trusted intermediaries.
The transition is opt-in. According to Figment's institutional staking analysis, the existing MEV-Boost relay infrastructure continues to function alongside the new ePBS system. No forced migration is required.
Flashbots has already pivoted in anticipation: the company launched BuilderNet in November 2024, migrating its centralized builders onto a decentralized system running in trusted execution environments. Its SUAVE chain was archived in May 2025 as the company refocused on BuilderNet and Flashnet infrastructure.
Ethereum currently executes transactions sequentially because nodes cannot predict which state a block will access until execution begins. EIP-7928 eliminates this bottleneck by requiring block producers to provide structured data describing each transaction's state access and resulting state changes.
With this information available upfront, execution clients can:
The practical effect: Ethereum can safely raise its gas limit from the current 60 million toward 200 million per block. Stańczak indicated at the Bankless Summit that the increase will be phased — 100 million initially, doubling to 200 million once ePBS stabilizes. This 3.3x capacity expansion would bring L1 throughput from roughly 20 TPS toward a theoretical 10,000 TPS ceiling.
Average L1 gas fees currently sit at 0.125–0.89 gwei as of mid-June 2026, translating to roughly $0.16–0.22 per transaction. At a 200 million gas limit, per-unit gas costs could fall an additional 78%, according to estimates cited by CoinMarketCap and CryptoBriefing. However, the relationship between capacity and fees is non-linear: increased throughput may attract additional demand, partially offsetting fee compression.
A 3.3x gas limit increase without cost adjustments would enable proportionally more state creation — new contracts, new storage slots, new account entries — all of which nodes must store indefinitely. Glamsterdam addresses this with targeted repricing:
| EIP | Mechanism | Purpose | |-----|-----------|---------| | EIP-8037 | Raises state-creation gas costs | Prevents excessive contract proliferation under higher limits | | EIP-7976 | Increases calldata floor cost | Primarily affects rollup batch posting economics | | EIP-7981 | Raises EIP-2930 access list costs | Recalibrates pricing post-BAL infrastructure | | EIP-7778 | Removes gas refunds from block accounting | Improves builder bid predictability |
An earlier proposal for dynamic per-state-byte pricing was abandoned in favor of the fixed-cost model in EIP-8037. The trade-off: simpler implementation at the cost of less granular state management. For rollups posting data via calldata (as opposed to blobs), EIP-7976's floor increase raises fallback posting costs, further incentivizing the migration to blob-based data availability.
MEV extraction on Ethereum totals an estimated $1–5 billion annually, according to webthreepedia's economic value framework. The current system concentrates this value in a small number of sophisticated builders who maintain private order flow agreements with searchers and protocols.
ePBS restructures this market in three ways:
Transparency: Builder bids and payload commitments become on-chain, visible events rather than opaque off-chain negotiations.
Competition: Solo validators can participate in block production without needing to trust a relay, lowering barriers to entry for smaller operators.
Timing: The expanded 9-second payload propagation window gives builders more time to construct optimal blocks, reducing the premium for low-latency infrastructure.
However, the "free option problem" persists. According to Bitfinex's analysis, proposers who commit to a builder's bid can, under certain conditions, choose to build their own block instead — an issue that affects approximately 0.82% of blocks on average, rising to 6% during volatile market periods. Whether ePBS fully solves this remains an open question; the PTC mechanism adds friction but does not eliminate the incentive entirely.
Glamsterdam does not displace Layer 2 networks. Even at 200 million gas per block, L2s remain cheaper for cost-sensitive activity and offer sub-second finality that L1 cannot match under 12-second slot times.
The upgrade's effects on L2 economics are indirect:
The likely outcome is a cleaner division of labor: L2s handle high-volume execution; L1 serves as the settlement and security anchor.
With 35.9 million ETH staked (28.9% of total supply) and average yields at 3.3%, Glamsterdam alters the institutional staking calculus in several areas:
Exit Liquidity: EIP-8080 and EIP-8061, under consideration for Glamsterdam, would materially reduce time-to-liquidity for large staked positions. The current staking queue contains 2.96 million ETH awaiting entry, while the exit queue has hit zero multiple times in 2026 — indicating relatively fluid exit conditions that the upgrade would further improve.
MEV Reward Distribution: A more competitive, trust-minimized builder market under ePBS could redistribute MEV revenue more broadly across validators. Figment, operating 1.48 million staked ETH (4.1% of supply), notes that "a more competitive, trust-minimized builder market could be positive for the variable execution layer component of staking rewards."
Hardware Requirements: Parallel execution via BALs requires node operators to audit storage IOPS capacity. PTC duties increase network latency sensitivity. Institutional operators running large validator fleets face non-trivial infrastructure upgrade costs.
Staking ETF Impact: BlackRock's iShares Staked Ethereum Trust ETF and Grayscale's staking ETHE (approximately $1.4 billion) pass staking yield through to traditional finance investors. Changes to reward composition — more predictable base rewards, restructured MEV distribution — directly affect fund performance metrics.
Three factors could push activation into Q4 2026 or later:
1. ePBS Complexity: Cross-client implementation of EIP-7732 remains the primary bottleneck. Multi-client devnets are running as of late May 2026, but stability under mainnet-scale load conditions is unproven.
2. Leadership Transitions: Protocol contributors Barnabé Monnot and Tim Beiko are departing the Ethereum Foundation, while Alex Stokes is taking a sabbatical. Responsibilities transfer to new Protocol Cluster leads Will Corcoran, Kev Wedderburn, and Fredrik Svantes — a transition occurring during active upgrade preparation.
3. Gas Repricing Calibration: Testing EIP-8037 and EIP-7976 at 200 million gas requires simulated mainnet loads that accurately model state growth and calldata demand. Miscalibration risks either insufficient state bloat protection or excessive cost increases for legitimate use cases.
Glamsterdam's predecessor Fusaka shipped in December 2025 and introduced PeerDAS. Stańczak previously cautioned against premature focus on Glamsterdam at the expense of Fusaka timelines — a warning that proved prescient as Fusaka itself experienced delays. The successor fork Hegotá, tentatively targeting Q4 2026 to Q1 2027, would introduce FOCIL for censorship resistance and potentially 6-second slots.
Glamsterdam is fundamentally an economic restructuring, not merely a performance upgrade. By enshrining proposer-builder separation into the protocol, Ethereum eliminates an entire layer of trusted intermediaries from its block production pipeline. By introducing block-level access lists, it enables the parallel execution needed to safely triple capacity.
The economic implications are significant. Ethereum L1 currently generates approximately $65 million in annualized fee revenue — a fraction of the $5–8 billion in total ecosystem value flows identified in webthreepedia's economic analysis. A 3.3x capacity increase at current demand levels would compress per-unit fees further, widening the gap between on-chain revenue and ecosystem costs. Whether increased throughput generates sufficient additional transaction volume to offset per-unit fee compression will determine Glamsterdam's net economic effect.
For the MEV market, ePBS represents a direct challenge to the current builder oligopoly. The transition is opt-in and gradual, but the direction is clear: Ethereum is internalizing block production economics that were previously captured by off-chain infrastructure providers. The $1–5 billion annual MEV extraction economy faces redistribution, not elimination.
The upgrade's success hinges on execution. Ethereum's track record on upgrade timelines is mixed — Fusaka slipped, and Glamsterdam has already moved from H1 to Q3 2026. With key protocol contributors departing and new leadership assuming responsibilities mid-cycle, the coordination challenge Stańczak warned about is now a live operational risk.