Ethereum's Glamsterdam hard fork — the network's largest protocol change since the September 2022 Merge — cleared its final devnet milestone in June 2026 and launched the Platåberget public testnet on August 17. Core developers proposed the Sepolia testnet fork for September 28 at epoch 351,232 d...
"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 Contributor
Ethereum's Glamsterdam hard fork — the network's largest protocol change since the September 2022 Merge — cleared its final devnet milestone in June 2026 and launched the Platåberget public testnet on August 17. Core developers proposed the Sepolia testnet fork for September 28 at epoch 351,232 during the August 20 All Core Devs Consensus call (ACDC #185), with no objections raised. Mainnet activation is targeted for Q4 2026.
The upgrade ships ten Ethereum Improvement Proposals under the Glamsterdam Meta EIP-7773. The two headliners — EIP-7732 (Enshrined Proposer-Builder Separation) and EIP-7928 (Block-Level Access Lists) — restructure how blocks are built and executed at the protocol level. The remaining eight EIPs adjust gas pricing, increase contract size limits, and introduce new opcodes. Combined, the changes clear the path toward a 200 million gas limit (up from ~60 million today), targeting roughly 10,000 TPS-equivalent throughput and an estimated 78.6% reduction in L1 transaction fees.
The fork arrives as approximately 90% of Ethereum blocks route through MEV-Boost relays, with just five builders controlling 96.7% of all MEV blocks. EIP-7732 aims to eliminate the trusted relay dependency that sustains this concentration. Whether the upgrade ships on schedule depends on client stability during the Sepolia and Hoodi testnet phases — multiple clients experienced issues around builder functionality in the most recent devnet round.
Glamsterdam follows Fusaka (activated December 2025) and precedes Hegotá, tentatively scheduled for H1 2027. The upgrade was originally targeted for H1 2026 but slipped as scope expanded beyond initial estimates.
Development milestones to date:
| Date | Milestone | |------|-----------| | Feb 4, 2026 | First Glamsterdam gas repricing working group call | | Jun 17, 2026 | EIP-7773 meta EIP locks in ten proposals; final devnet phase begins | | Aug 17, 2026 | Platåberget public testnet launches | | Aug 20, 2026 | Glamsterdam fork activates on Platåberget | | Aug 20, 2026 | ACDC #185 proposes Sepolia fork at epoch 351,232 (Sep 28, 2026) | | Sep 28, 2026 (proposed) | Sepolia testnet fork at 14:44:48 UTC | | Q4 2026 (target) | Mainnet activation |
The Ethereum Foundation noted on August 24 that a non-finality devnet will follow in approximately one month to test consensus edge cases. Sepolia and Hoodi testnets must complete the upgrade before mainnet deployment proceeds. Client issues during recent devnet rounds — where multiple Ethereum clients stalled around new builder functionality — remain a risk factor for the September 28 Sepolia date.
EIP-7773 tracks the following ten proposals as Scheduled for Inclusion:
| EIP | Name | Category | |-----|------|----------| | EIP-7732 | Enshrined Proposer-Builder Separation (ePBS) | Consensus / Block Construction | | EIP-7928 | Block-Level Access Lists (BALs) | Execution / Parallelization | | EIP-7976 | Increase Calldata Floor Cost | Gas Repricing | | EIP-7981 | Increase Access List Cost | Gas Repricing | | EIP-8037 | State Creation Gas Cost Increase | Gas Repricing | | EIP-7954 | Increase Maximum Contract Size | EVM Expansion | | EIP-7778 | Block Gas Accounting Without Refunds | Gas Mechanics | | EIP-8024 | Backward-Compatible SWAPN, DUPN, EXCHANGE | EVM Opcodes | | EIP-7843 | SLOTNUM Opcode | EVM Opcodes | | EIP-7708 | ETH Transfers Emit a Log | Execution Logging |
One notable absence: EIP-7782 (6-second slots) was shelved after developers deemed it insufficiently mature for inclusion, with inadequate client implementation progress.
The problem. Today, roughly 90% of Ethereum blocks are produced through MEV-Boost, an off-protocol relay system. According to relay data from relayscan.io (August 23, 2026), market share among relays is as follows: Ultra Sound at 26.39%, bloXroute Max Profit at 22.18%, Titan Relay at 21.45%, bloXroute Regulated at 19.00%, Aestus at 6.66%, Flashbots at 2.35%, and Agnostic Relay at 1.40%. On the builder side, concentration is more extreme: Titan builder controls 55.78% of blocks, Quasar holds 19.71%, and BuilderNet accounts for 14.36%. Five builders together produce 96.7% of all MEV blocks.
This concentration creates a trust dependency on off-chain relays and centralizes block production among a small number of entities — a structural vulnerability that contradicts Ethereum's decentralization goals.
The fix. EIP-7732 moves the proposer-builder separation mechanism directly into the consensus protocol. Under the new design, proposers select execution proposers (builders) on-chain, eliminating the need for trusted third-party relays. The proposal introduces a Payload Timeliness Committee (PTC), a new validator duty that attests payloads arrive on schedule, with protocol-enforced recourse if a builder fails to deliver. The propagation window extends from approximately 2 seconds to roughly 9 seconds to accommodate the new two-phase commit structure.
Proponents estimate ePBS could reduce MEV losses by up to 70%, according to analysis cited by Everstake and thirdweb. MEV-Boost revenue currently adds 10–30% to validator rewards on top of a 2.78% base APR, translating to roughly 0.28–0.83% additional yield. How ePBS reshapes this yield distribution remains an open question.
The problem. Ethereum processes transactions sequentially within each block. This is a throughput ceiling: even as hardware improves, blocks cannot execute faster than the serial processing time allows.
The fix. EIP-7928 introduces Block-Level Access Lists (BALs), which record all accounts and storage locations accessed during block execution along with post-execution values. This metadata allows clients to safely execute transactions in parallel when they touch disjoint state.
Historical data analyzed by the EIP's authors shows 60–80% of transactions access disjoint storage slots, enabling effective parallelization. The remaining 20–40% can still benefit from parallel processing through post-transaction state diffs. Average BAL size is approximately 70 KiB per block.
A new block_access_list_hash field is added to the block header containing the Keccak-256 hash of the RLP-encoded access list. This is not backward-compatible and requires the hard fork.
The combination of BALs and the gas repricing package clears the engineering path toward a 200 million gas limit — roughly 3.3x the current ~60 million — targeting approximately 10,000 TPS-equivalent throughput under realistic workloads. The 200 million figure is a design target, not a value the fork enforces; validators would step the limit up via standard gas-vote signaling as node performance under larger blocks is validated.
Five of the ten Glamsterdam EIPs adjust gas costs to reflect actual computational resource consumption on modern hardware. Current gas prices for many operations were last set during the Berlin fork in April 2021.
EIP-7976 increases the calldata floor cost, discouraging calldata-heavy transactions that consume bandwidth disproportionate to their gas expenditure. EIP-7981 raises EIP-2930 access list entry costs. EIP-8037 increases gas costs for state-creating operations (new accounts, storage slots, bytecode deployment), better reflecting the long-term storage burden these operations impose on the network. EIP-7778 removes gas refunds from block-level accounting, simplifying gas mechanics. EIP-7954 increases the maximum contract size, expanding the design space for on-chain applications.
The net effect of the repricing package is estimated at approximately 78.6% lower L1 fees under the new gas schedule, according to analysis cited by multiple sources including IG and Phemex. The Ethereum Foundation warned in an August 24 blog post that a "small set of smart contracts" relying on hardcoded gas assumptions — fixed stipends, hardcoded gas values in calls, logic branching on gasleft(), or presigned transactions with fixed limits — may break without updates. The Foundation stated it has begun direct outreach to the most-affected builders. The majority of smart contracts are expected to be unaffected.
Glamsterdam does not abandon Ethereum's rollup-centric roadmap. Instead, it strengthens L1 as a settlement layer while reducing costs for rollups posting data to mainnet.
The predecessor upgrade, EIP-4844 (Proto-Danksharding, activated March 2024), cut L2 transaction costs by 90–99%, dropping fees from $0.50–$5.00 to $0.001–$0.05. Glamsterdam's gas limit expansion and blob capacity increase will push L2 fees lower still, while higher L1 throughput reduces congestion during demand spikes.
For the L2 duopoly that has emerged — as noted in prior webthreepedia analysis — cheaper L1 settlement costs reduce one of the primary value-capture mechanisms for rollup sequencers: the spread between user fees and L1 posting costs. This compression may accelerate fee competition among L2s.
From an economic value distribution perspective, the shift has implications for how protocol revenue flows. Enshrining PBS on-chain redirects value that currently accrues to off-chain relay operators back into the protocol layer. The gas repricing simultaneously reduces the absolute amount of fee revenue flowing through the system, even as throughput increases. Whether higher volume offsets lower per-transaction fees — a dynamic familiar to any toll-road operator — will determine whether Ethereum's on-chain revenue improves or deteriorates post-Glamsterdam.
Client stability. Multiple Ethereum clients experienced stalls and slowdowns around builder functionality during the most recent devnet round. If these issues persist on Sepolia, the mainnet timeline could slip into 2027.
Validator hardware requirements. A 200 million gas limit triples the computational load per block. Node operators running minimal hardware may face degraded propagation or attestation miss rates. The gas limit increase is not enforced by the fork but depends on voluntary validator signaling — a coordination exercise that could move faster or slower than developers anticipate.
MEV redistribution. Eliminating trusted relays does not eliminate MEV. It redirects it. How value redistributes among proposers, builders, and the protocol under ePBS is not yet empirically tested at scale. The 70% MEV loss reduction figure cited by proponents is a model estimate, not an observed outcome.
Smart contract breakage. The Ethereum Foundation's August 24 warning about hardcoded gas assumptions is not cosmetic. Any contract or transaction pipeline that assumes current gas costs without room for adjustment faces potential failure. The Foundation describes the affected set as "small," but the DeFi ecosystem's composability means a single broken contract in a dependency chain can have outsized effects.
Scope creep history. Glamsterdam has already slipped from H1 2026 to Q4 2026 as scope expanded. The shelving of EIP-7782 (6-second slots) suggests developers have trimmed where possible, but the ten-EIP bundle is still the most complex single upgrade since the Merge.
Glamsterdam represents Ethereum's attempt to address two structural problems simultaneously: the centralization of block production through off-chain relays and the serial execution bottleneck that constrains L1 throughput. The ten-EIP bundle is technically ambitious — the largest protocol change in four years — and the testing runway from Platåberget through Sepolia to mainnet leaves limited margin for client instability.
The economic implications are material. Enshrining PBS on-chain redistributes value currently captured by relay operators. Gas repricing reduces per-transaction cost but may also compress protocol revenue unless volume scales proportionally. For validators, the introduction of the Payload Timeliness Committee creates a new operational duty that solo stakers and institutional operators alike must prepare for.
The upgrade does not ship higher throughput automatically. A 200 million gas limit requires validator consensus via gas-vote signaling, client optimization, and empirical proof that block propagation holds under larger payloads. Whether Glamsterdam delivers on its 10,000 TPS target or produces something more modest will only become clear once mainnet has run under load.
Ethereum's core developers have signaled this is the beginning of a broader restructuring. Vitalik Buterin described the network's current trajectory as its "biggest rebuild since the Merge," with the Lean Ethereum roadmap calling for quantum resistance, privacy, and a move beyond the current EVM over the next three to four years. Glamsterdam is the first concrete delivery against that vision.