Ethereum's largest protocol upgrade since the September 2022 Merge has slipped its mainnet target for the third time since June 2026. Glamsterdam — which bundles enshrined proposer-builder separation (ePBS), block-level access lists, and a gas-limit pathway from 60 million to 200 million — now ta...
"Glamsterdam still needs a stable devnet." — Christine D. Kim, Founder, Protocol Watch Research & Advisory
Ethereum's largest protocol upgrade since the September 2022 Merge has slipped its mainnet target for the third time since June 2026. Glamsterdam — which bundles enshrined proposer-builder separation (ePBS), block-level access lists, and a gas-limit pathway from 60 million to 200 million — now targets a Sepolia testnet fork on October 6 at 13:53 UTC, contingent on a clean run of Devnet-11, which launches today, September 14, with 84,000 validators. The mainnet window, originally June, has receded to December 2026 at the earliest.
The delays stem from persistent finality failures across nine sequential devnets, a consensus-layer bug on Devnet-8 that could have halted the live network, and EIP-8037 implementation discrepancies that forced all execution-client teams to revise their code. Devnet-9, running since September 1 with 1,000 validators, still cannot finalize blocks normally. If Devnet-11 reproduces that failure at 84x the validator count, the October 6 Sepolia date reverts to the core-developer calls for rescheduling.
The stakes are structural. Glamsterdam rewrites how Ethereum creates, verifies, and propagates blocks. It moves MEV infrastructure from off-protocol relays into consensus rules, enables parallel transaction execution for the first time on L1, and introduces a state-growth pricing model designed to cap database expansion at 120 GiB per year under a 200-million gas limit — roughly triple the current L1 capacity.
Glamsterdam's original mainnet target was June 2026. It was pushed to Q3, then Q4, and now hinges on a Sepolia testnet fork that itself has slipped from August 3 to October 6.
| Milestone | Original Date | Current Date | Status | |---|---|---|---| | Devnet-0 through Devnet-9 | March–Sept 2026 | Completed | Nine sequential devnets; none achieved stable finality | | Devnet-11 | — | Sept 14, 2026 | Launching today; 84,000 validators | | Devnet-11 Gloas fork | — | Sept 16, 2026 | First activation test on Devnet-11 | | Sepolia testnet fork | Aug 3, 2026 | Oct 6, 2026 | Conditional on Devnet-11 stability | | Hoodi testnet fork | TBD | TBD | Not scheduled; awaits Sepolia success | | Mainnet | June 2026 | Dec 2026 (earliest) | Conditional |
Developers skipped Devnet-10 entirely, jumping from Devnet-9 to Devnet-11 after determining that the scope of fixes warranted a fresh configuration rather than an incremental patch. The decision was confirmed at the September 3 All Core Developers Consensus (ACDC) call #186.
The slippage pattern reflects Ethereum's post-Fusaka commitment to faster upgrade cycles running into the engineering reality of shipping the protocol's most complex single change. According to a CoinEdition analysis, this is the first major test of that accelerated development pace.
Glamsterdam combines a Prague execution-layer fork with consensus-layer changes in a single release containing at least 12 EIPs (tracked under meta-EIP 7773). The three headline changes:
1. Enshrined Proposer-Builder Separation (EIP-7732): Moves block-building roles from off-protocol relays into Ethereum's consensus rules, making MEV-Boost optional rather than load-bearing.
2. Block-Level Access Lists (EIP-7928): Creates per-block maps of all accounts and storage slots touched by transactions, enabling parallel execution of non-conflicting operations. This is a prerequisite for future gas-limit increases toward 200 million.
3. State Creation Gas Cost Increase (EIP-8037): Introduces a fixed cost-per-state-byte (CPSB) model with a dedicated state gas reservoir, targeting 120 GiB/year maximum database growth at the 200M gas limit.
Additional EIPs include EIP-2780 (reducing intrinsic transaction gas — plain ETH transfers may drop to 15,000 gas for zero-value transactions), EIP-7708 (ETH transfers emit ERC-20-style logs), EIP-7954 (contract code size limit rises from 24 KiB to 64 KiB), EIP-8038 (state-access gas cost updates for DoS resilience), EIP-8045 (exclusion of slashed validators from block proposals), and EIP-8061 (separate churn lanes for exits and consolidations, increasing exit capacity approximately 4x).
Finality — the point at which a block becomes irreversible — is Ethereum's core safety guarantee. Every devnet through Devnet-9 has failed to achieve stable finality under the Glamsterdam rule set.
Devnet-8 exposed a consensus-layer bug where a block could be created with an identical hash to its parent, a condition that Ethereum Foundation engineer Stefan Starflinger noted could cause a network-wide interruption on mainnet. Devnet-9, launched September 1 with 1,000 validators, patched that bug but introduced new finality failures tied to EIP-8037's split implementation across execution and consensus layers.
The EIP-8037 issue is particularly instructive. State-gas repricing requires coordinated changes to both the execution layer (which prices gas) and the consensus layer (which validates blocks against those prices). The initial specification did not fully account for the interaction between EIP-8037 and EIP-7251's MaxEB validator consolidation (shipped in Pectra), forcing all execution-client teams to revise their state-creation gas logic.
Devnet-11, launching September 14 with 84,000 validators — an 84x increase over Devnet-9 — represents the closest approximation to mainnet conditions tested to date. Ethereum's mainnet runs approximately 910,000 active validators staking 43.1 million ETH (35.3% of total supply), according to beaconcha.in data. Lido and Optimism have requested at least one stable day of operation before fork activation proceeds.
Maximal extractable value (MEV) — the profit validators and builders earn by reordering transactions — currently flows through off-protocol middleware. MEV-Boost, the dominant relay architecture, processes the majority of Ethereum blocks. As of late 2025, relay market share was fragmented: Ultra Sound relay handled 32.3% of MEV-Boost payloads, Titan Relay 24.75%, bloXroute relays a combined 26%, and Flashbots' own relay 3.44%, according to relay monitoring data.
EIP-7732 writes proposer-builder separation into Ethereum's consensus rules. Under ePBS:
The trust assumption currently sitting with third-party relays moves into the protocol itself. Relays become optional infrastructure rather than critical middleware. For stakers, block-selection rights increase and MEV rewards become more predictable. For relay operators, the change is existential: their primary function gets absorbed by the base protocol.
The engineering difficulty of ePBS lies in handling two-party coordination and disagreement inside consensus rather than outside it. According to analysis from bex.co, the implementation interacts with every piece of the post-Pectra stack in ways the original specifications did not fully anticipate.
Ethereum's current block gas limit sits at 60 million, set after the December 2025 Fusaka upgrade. Glamsterdam establishes the technical conditions for validators to raise it toward 200 million — a roughly 3.3x increase in L1 throughput, estimated at up to 10,000 TPS-equivalent under realistic workloads.
The constraint is state growth. Every new account, contract, and storage slot permanently enlarges Ethereum's state database. Without repricing, a 200M gas limit would accelerate state bloat to unsustainable levels for node operators.
EIP-8037 addresses this through three mechanisms:
The Ethereum Foundation has warned that the 21,000-gas assumption for standard ETH transfers — hardcoded into wallets, exchanges, and custody platforms for years — will break under Glamsterdam. EIP-2780 restructures intrinsic gas into measured components. While a standard transfer remains near 21,000 gas, zero-value and self-transfer transactions drop to 15,000 and 12,000 gas respectively. Any system that hardcodes the 21,000 figure requires updates.
Glamsterdam is shipping under new protocol leadership. Tim Beiko and Barnabé Monnot departed, and Alex Stokes is on sabbatical. Three new Protocol cluster leads — Will Corcoran, Kev Wedderburn, and Fredrik — have assumed responsibility for shepherding the upgrade through its final testing phases.
The leadership transition coincides with what is, by several measures, the most complex single upgrade in Ethereum's history. The combination of new management and persistent technical failures has extended the timeline but, according to developers, has not changed the scope. No EIPs have been removed from the Glamsterdam specification.
The Ethereum roadmap names two subsequent upgrades: Hegotá, currently in scoping, and a longer-term initiative called Strawmap focused on quantum-resistant cryptography. The December 2025 Fusaka upgrade established a faster release cadence. Glamsterdam is the first test of whether that cadence can absorb a structurally complex upgrade without indefinite delays.
If Glamsterdam ships in Q4 2026, Ethereum will have completed two major upgrades in roughly 12 months — the pace developers committed to. If it slips into 2027, the faster-cycle thesis faces questions about whether it can accommodate fundamental protocol changes alongside incremental ones.
Glamsterdam's repeated delays reflect a structural tension in Ethereum's development model: the commitment to faster release cycles colliding with the engineering complexity of rewriting block production, transaction execution, and state pricing simultaneously. The upgrade touches every layer of the stack — consensus, execution, networking — and its EIPs interact with each other and with Pectra's changes in ways that surfaced only under devnet testing.
The October 6 Sepolia date is not a deadline; it is a conditional target. If Devnet-11's 84,000 validators finalize blocks and survive the September 16 Gloas fork without the failures that plagued Devnets 8 and 9, the path to a December mainnet activation remains plausible. If not, the timeline resets to 2027, and Ethereum's faster-upgrade thesis faces its first public failure.
The market implication is not about ETH price. It is about infrastructure readiness. Every wallet, exchange, custody platform, and L2 rollup that hardcodes gas assumptions, relies on MEV-Boost relays as critical infrastructure, or runs validators with current client configurations will require updates. The scope of downstream preparation is proportional to the scope of the protocol change — and Glamsterdam's scope is, by any objective measure, large.