Ethereum's Glamsterdam hard fork — the network's largest protocol change since the Merge — has locked its EIP bundle and entered public testnet preparation, with the Sepolia fork provisionally scheduled for September 28, 2026. The upgrade's two headline proposals, EIP-7732 (enshrined Proposer-Bui...
"The protocol is approaching its endgame, yet we've only just begun scratching the surface of what a permissionless, scalable, cryptoeconomically secure, and cheaply verifiable world computer can do." — Tim Beiko, Ethereum Foundation (Protocol Team)
Ethereum's Glamsterdam hard fork — the network's largest protocol change since the Merge — has locked its EIP bundle and entered public testnet preparation, with the Sepolia fork provisionally scheduled for September 28, 2026. The upgrade's two headline proposals, EIP-7732 (enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists), aim to triple the network's execution-layer capacity by clearing a path toward a 200 million gas-limit floor, up from the current ~60 million.
The timing is not coincidental. Ethereum's gross fee revenue fell 69.3% year-over-year in H1 2026 to $127 million, according to 21Shares research, as Layer 2 rollups captured an increasing share of transaction volume and economic value. Glamsterdam represents Ethereum's structural response: reclaim throughput at the base layer while reducing dependence on off-chain infrastructure that has concentrated block production among a small number of actors.
Glamsterdam combines two coordinated forks: Gloas on the consensus layer and Amsterdam on the execution layer. The upgrade entered its final devnet phase (devnet-6) in mid-2026 after the Soldøgn interop event concluded on May 2, 2026, where six client teams verified cross-client compatibility.
Key milestones and dates:
| Milestone | Date / Status | |---|---| | Soldøgn interop devnet | Concluded May 2, 2026 | | Devnet-6 (final) | Active as of August 2026 | | Sepolia testnet fork | Provisionally Sept. 28, 2026 (14:44:48 UTC) | | Holesky/Hoodi testnet forks | Not yet scheduled | | Mainnet activation | Realistic window: Q4 2026 |
The Sepolia date was pinned during the August 20, 2026 All Core Developers call, where six client teams confirmed the epoch and slot without dissent. However, developers noted the date remains provisional. Given that recent forks (Dencun, Pectra, Fusaka) required two to four months of public-testnet seasoning, a Q4 2026 mainnet window is the firmer estimate. The upgrade was originally targeted for June 2026 but slipped due to implementation complexity.
Leadership changes add a personnel dimension. Tim Beiko and Barnabé Monnot have transitioned from active protocol coordination. Will Corcoran, Kev Wedderburn, and a developer identified as Fredrik now lead the Protocol Team. Alex Stokes has taken a sabbatical.
The most consequential change in Glamsterdam. EIP-7732 moves block building into Ethereum's consensus rules, replacing the off-chain MEV-Boost relay system that currently mediates between validators and specialized block builders.
Under the current system, a validator who wants to maximize revenue outsources block construction to a builder via MEV-Boost. The builder assembles a block that extracts maximum value (via arbitrage, liquidations, and sandwich attacks), bids for the right to fill the block, and submits it through a trusted relay. The validator accepts the highest bid without seeing the block's contents.
EIP-7732 replaces this with an in-protocol mechanism:
This eliminates the trust assumption on relays and creates a protocol-level market for block construction.
EIP-7928 introduces a data structure that maps every account and storage slot accessed during block execution, along with post-execution values. This enables:
BALs are constrained by the block gas limit rather than a fixed item cap. According to core developers, this is the prerequisite that makes a 200 million gas-limit floor technically sustainable. Without BALs, sequential execution becomes the bottleneck long before gas limits reach that level.
The upgrade bundles several additional proposals:
| EIP | Function | |---|---| | EIP-2780 | Restructures the flat 21,000 gas base cost into measured resource components. Zero-value transactions drop to ~15,000 gas; self-transfers to ~12,000 | | EIP-8037 | Increases state creation gas costs to limit database growth under higher gas limits | | EIP-7954 | Raises maximum contract code size from 24 KiB to 64 KiB; initcode limit from 48 KiB to 128 KiB | | EIP-7708 | ETH transfers now emit logs matching the ERC-20 Transfer event format | | EIP-7778 | Removes gas refunds from block-level accounting | | EIP-7843 | Adds SLOTNUM opcode | | EIP-8024 | Backward-compatible SWAPN, DUPN, EXCHANGE stack opcodes | | EIP-7976 | Increases calldata floor cost | | EIP-7981 | Increases access list cost |
The economic case for ePBS is rooted in concentration data. As of mid-2026, three relays route approximately 88% of all MEV-Boost blocks. The top two builders — Titan and Quasar — construct roughly 73% of all blocks. Including the third and fourth largest (Eureka and BuilderNet) brings that figure above 95%.
This creates a structural dependency: Ethereum's block production pipeline runs through a handful of off-chain intermediaries that operate outside protocol governance. A relay outage or censorship decision by two or three entities could affect the majority of Ethereum blocks.
Flashbots launched BuilderNet, a decentralized block-building network using Trusted Execution Environments (TEEs), as a partial mitigation. But ePBS is the protocol-level solution. By moving builder identity, bid submission, and payment enforcement into consensus, EIP-7732 eliminates the relay as a single point of failure and trust.
The economic implication: proponents estimate ePBS could reduce MEV losses borne by ordinary users by up to 70%, though this figure depends on builder competition dynamics that have not yet been tested at scale.
The gas repricing bundle (anchored by EIP-2780) replaces Ethereum's flat 21,000 gas intrinsic transaction cost — unchanged since 2015 — with resource-based pricing. Under the new model:
EIP-8037 counterbalances this by increasing the gas cost of state creation operations. The logic: cheaper transactions are fine, but new state entries that permanently enlarge the database should cost more. This creates a pricing structure that encourages transaction throughput while discouraging state bloat — a necessary precondition for raising the gas limit.
The design target is approximately 10,000 TPS-equivalent throughput under realistic workloads at the 200 million gas-limit floor, though developers have been careful to frame this as a ceiling, not a guarantee.
Glamsterdam arrives against a backdrop of deteriorating L1 economics. Data from 21Shares:
| Metric | H1 2025 | H1 2026 | Change | |---|---|---|---| | Gross revenue | $414M | $127M | -69.3% | | Fee revenue vs. 2021 peak | — | — | -95%+ |
The decline affected all four fee categories: base fees, Layer 2 data fees, priority fees, and MEV. EIP-4844's blob fee market, introduced in March 2024, slashed L2 data costs by over 90%, accelerating migration but also reducing L1 fee capture.
However, usage metrics tell a different story. In H1 2026, monthly active addresses grew 15% year-over-year to 8.4 million. Smart contract deployments increased 74% to over 1.3 million. Stablecoins on Ethereum grew 22% to approximately $156 billion. Ethereum commands roughly $16 billion of the $34 billion in tokenized real-world assets on-chain.
The divergence — rising usage, falling revenue — is the core economic problem Glamsterdam addresses. By tripling L1 capacity and reducing per-transaction costs, the upgrade bets that lower prices will drive sufficient volume increases to stabilize or grow aggregate fee revenue. Whether this works depends on demand elasticity that remains untested at these throughput levels.
Two proposals were deferred to Hegotá, the upgrade following Glamsterdam (targeted for H1 2027):
EIP-7782 (6-second slots): Would have halved Ethereum's slot time from 12 seconds to 6 seconds, doubling block frequency. Deferred due to implementation immaturity and conflicts with ZK-proving workflows that depend on current timing assumptions.
FOCIL (Fork-Choice Inclusion Lists): Would have given proposers the ability to mandate that certain transactions be included in a block, serving as a censorship resistance mechanism. Moved to Hegotá to reduce Glamsterdam's implementation complexity.
The exclusions are notable because both address censorship resistance — a stated priority for the Ethereum community. Their deferral suggests that the developer team prioritized throughput and MEV infrastructure over censorship tooling for this upgrade cycle.
Validators and stakers must update both Consensus Layer (CL) and Execution Layer (EL) clients before the fork activates. The new PTC attestation duty introduced by ePBS adds a responsibility: selected validators must attest to whether a builder's payload arrived within the timing window. Failure to attest may result in minor penalties.
Node operators face a more significant operational change. BALs enable parallel disk reads, but this means storage I/O performance becomes a more critical bottleneck. Operators should audit storage IOPS capacity and optimize for concurrent read patterns. Nodes running on consumer-grade hardware with slow disk I/O may struggle under higher gas limits.
ETH holders face no required action. The upgrade does not change staking mechanics, token supply dynamics, or the burn mechanism introduced by EIP-1559.
Glamsterdam is Ethereum's most significant execution-layer upgrade since the Merge. It addresses two structural weaknesses simultaneously: the concentration of block production in off-chain relays and the sequential execution bottleneck that caps throughput at current gas limits.
The economic stakes are clear. L1 fee revenue has fallen 69.3% year-over-year while the network's usage metrics continue to grow — a divergence that, if unaddressed, undermines Ethereum's value capture relative to Layer 2 networks that sit atop it. Whether tripling capacity reverses this trend depends on demand elasticity assumptions that have not been tested at scale.
The technical risks are implementation complexity (the upgrade already slipped from its June target) and the untested behavior of ePBS under adversarial market conditions. The personnel changes at the Protocol Team add execution uncertainty.
What is not in dispute: the EIP bundle is locked, the testnet schedule is set, and Ethereum is committing to a structural shift in how its base layer processes transactions, builds blocks, and prices resources.