Ethereum developers cancelled the Glamsterdam hard fork activation on the Hoodi testnet on October 22, three days before its scheduled October 26 deployment at 17:42:48 UTC. The cancellation followed a ChainSecurity audit that identified 13 deployed smart contracts at risk of behavioral changes u...
"We found 13 cases where the interaction between EIP-8037 and EIP-8038 could turn previously retryable contract failures into permanent, irreversible outcomes." — ChainSecurity, Ethereum Smart Contract Auditor
Ethereum developers cancelled the Glamsterdam hard fork activation on the Hoodi testnet on October 22, three days before its scheduled October 26 deployment at 17:42:48 UTC. The cancellation followed a ChainSecurity audit that identified 13 deployed smart contracts at risk of behavioral changes under proposed gas-repricing rules in EIP-8037 and EIP-8038. Node operators have been instructed to run client software that excludes the scheduled fork.
The postponement puts Ethereum's tentative December mainnet window in doubt — the third timeline slip for Glamsterdam, which was originally targeted for H1 2026. ETH fell 3.5% on October 22, the day of the announcement, closing at $2,386 from $2,472 the prior session. As of October 26, ETH trades near $2,320, down 11% for the month, with $44 million in long liquidations and $1.18 billion in October ETF outflows adding to downward pressure.
Glamsterdam carries two of the most consequential protocol changes in Ethereum's post-Merge history: Enshrined Proposer-Builder Separation (EIP-7732) and Block-Level Access Lists (EIP-7928). Together, they would restructure how blocks are built, validated, and processed — a prerequisite for pushing the gas limit from 60 million to 200 million. The delay does not affect the technical merits of those proposals. It reflects the difficulty of repricing gas on a network with $80 billion in DeFi deposits and thousands of immutable contracts that cannot be patched.
On October 22, the Ethereum Foundation announced the cancellation of the Glamsterdam fork on the Hoodi testnet, which had been confirmed for October 26 at epoch 17:42:48 UTC. The planned activation would have also raised Hoodi's gas limit to 200 million, a critical stress test for the mainnet deployment.
The decision followed a security disclosure from ChainSecurity, published October 13, which flagged risks in the interaction between two gas-repricing EIPs: EIP-8037 (State Creation Gas Cost Increase) and EIP-8038 (State-Access Gas Cost Update). ChainSecurity identified 13 real-world smart contract deployments that would exhibit changed behavior under the new gas accounting rules.
Developers agreed to investigate the interactions before proceeding. No new Hoodi or mainnet activation date has been set. Previously suggested mainnet dates of December 1, 8, and 9 are now considered unreliable.
The Glamsterdam upgrade had already cleared its Sepolia testnet activation on October 6 without incident. However, Sepolia carries a fraction of the contract diversity found on mainnet, and the problematic interactions were only surfaced during deeper analysis of mainnet-deployed bytecode.
The two proposals are complementary gas-repricing mechanisms designed to align Ethereum's gas costs with the actual computational expense of state operations on a network whose state database has grown substantially since the last repricing.
EIP-8037: State Creation Gas Cost Increase introduces a new unit called CPSB (cost per state byte) set at 1,530 gas, targeting an average state growth of 120 GiB per year at a 150-million gas limit. It sets explicit costs per operation type: 64 bytes per storage slot, 120 bytes per new account, and 23 bytes per authorization. The proposal also introduces independent metering for state creation, separating it from the single transaction gas limit to allow larger contract deployments.
EIP-8038: State-Access Gas Cost Update raises the cold account access cost from 2,600 to 3,000 gas (+15%) and introduces explicit write surcharges for account and storage writes. It dissects state-touching operations into three cost components: access, write, and creation.
The conflict: When both EIPs interact, a failed external call can leave more remaining gas than the same operation would have under current rules. Contracts that use remaining gas to determine whether an operation failed permanently or should be retried can misinterpret the result. According to ChainSecurity, this can cause a calling contract to "record an unsuccessful operation without allowing users to retry it" — converting what was previously a recoverable failure into a permanent one.
The 13 affected contracts have not been publicly identified. Responsible disclosure is underway. The scope of potential impact remains under investigation.
Glamsterdam is governed by EIP-7773, the fork meta-EIP, which listed 19 proposals as of August 16, 2026 (Draft status). The upgrade combines two layers: Amsterdam (execution layer) and Gloas (consensus layer).
Headline proposals:
EIP-7732: Enshrined Proposer-Builder Separation (ePBS). The single largest architectural change since the Merge. Currently, block proposers outsource block construction to specialized builders through off-chain relay infrastructure — predominantly MEV-Boost relays. EIP-7732 moves this separation into the consensus protocol. A proposer commits to a builder's payload; the builder subsequently reveals it; the protocol settles payment on-chain. This eliminates trust assumptions in relay middleware. The proposal introduces the Payload Timeliness Committee (PTC), a new validator committee that attests to timely payload delivery. It expands the data propagation window from approximately 2 seconds to 9 seconds, a prerequisite for safely increasing throughput.
EIP-7928: Block-Level Access Lists (BALs). Creates enforced lists of accounts and storage locations accessed during a block, with post-transaction state changes. With this map available before execution, clients can prefetch state in parallel and process non-overlapping transactions concurrently instead of sequentially. This is the primary mechanism for pushing the gas limit from 60 million to 200 million.
Supporting proposals include:
Why ePBS matters for the relay market: Active MEV-Boost relays have dropped from 12 to 6 in 2026. Flashbots' own relay carries just 2.4-2.7% of daily payloads, down from post-Merge dominance. The builder market is concentrated: Titan Builder holds 47-55% market share, Quasar approximately 19%, BuilderNet 13%, and Eureka 11%. According to Flashbots' own August 2026 disclosure, builder profit per block has fallen 68% and relay operating margins are negative. ePBS would remove the relay trust layer entirely, but builder concentration is expected to persist until additional mechanisms like inclusion lists or encrypted mempools are deployed.
Glamsterdam's postponement follows a consistent pattern in Ethereum upgrade history:
| Upgrade | Original Target | Actual Deployment | Slip | |---------|----------------|-------------------|------| | The Merge | Q1 2022 | September 2022 | ~6 months | | Dencun | Q3 2023 | March 2024 | ~6 months | | Fusaka | November 2025 | December 3, 2025 | ~1 month | | Glamsterdam | H1 2026 | TBD (was Q4 2026) | 6+ months and counting |
The Ethereum Foundation announced in late 2025 its intention to shift to a twice-a-year hard fork cadence, beginning with Fusaka. Glamsterdam was meant to validate this new pace. Its slippage from H1 to Q4 2026, and now potentially into 2027, tests whether the accelerated cadence is sustainable for upgrades of this complexity.
Fusaka, by comparison, shipped with a relatively contained scope: PeerDAS (peer data availability sampling) as the headline feature, a gas limit increase to 60 million, and blob parameter updates. Glamsterdam carries two architectural overhauls — ePBS and BALs — alongside a comprehensive gas repricing. The scope difference may explain the divergent slippage profiles.
The Platåberget public testnet launched August 17 with a Glamsterdam fork on August 20. Sepolia activated October 6. The Hoodi activation was the final pre-mainnet gate. Its cancellation means the entire testnet progression must partially restart once the EIP-8037/8038 issues are resolved or the proposals are modified.
ETH price reaction to the cancellation was measurable but contained:
Contributing factors beyond the Glamsterdam delay include $1.18 billion in Ethereum ETF outflows during October, $44 million in ETH long liquidations, and broader risk-off positioning ahead of the Federal Reserve's upcoming meeting.
Ethereum ETFs have now recorded net outflows in five of the last six weeks, per data reported by CryptoTimes. The staking gap — the inability of U.S.-listed spot Ethereum ETFs to earn staking yield — continues to drive structural underperformance relative to direct ETH holdings, a factor this publication has previously identified.
The delay does not fundamentally alter Ethereum's value proposition. It does, however, extend the window during which the network operates without enshrined PBS, leaving the relay infrastructure in its current fragile state, and without parallel transaction processing, keeping throughput constrained at the 60-million gas limit set by Fusaka.
Three paths are available to developers:
Modify EIP-8037/8038 to eliminate the gas underflow interactions. This requires re-auditing and new testnet deployments, likely adding 8-12 weeks to the timeline.
Remove EIP-8037/8038 from Glamsterdam and ship ePBS and BALs without the gas repricing. This preserves the December window but defers the gas cost realignment that developers consider necessary for sustainable state growth management.
Delay the entire upgrade until all proposals are fully audited. This is the most conservative path and the one the development community has historically favored.
Signals to watch, according to the Ethereum Foundation: a rescheduled Hoodi epoch number and fresh client releases from teams including Lodestar and Nimbus. No timeline for these has been communicated.
The next major milestone in Ethereum's roadmap after Glamsterdam is Hegota, which would bring further L1 scaling. However, its scope and timeline depend on Glamsterdam shipping first.
Glamsterdam's cancellation is not a failure of engineering judgment. It is an illustration of the constraints Ethereum faces as a $280 billion network with immutable deployed contracts. Repricing gas — a necessary adjustment as state grows — cannot be done safely without exhaustive analysis of how existing contracts interpret gas accounting. ChainSecurity's identification of 13 vulnerable deployments, with responsible disclosure still underway, suggests the actual surface area may be larger.
The practical consequence is a delay in two upgrades that address real, measurable problems: relay fragility (ePBS) and throughput limitations (BALs). Every week of delay is a week the network runs on middleware that its own builders describe as economically unsustainable, and a week L2s operate with a 60-million gas limit instead of 200 million.
Ethereum's development process has consistently prioritized security over speed. The Merge slipped six months. Dencun slipped six months. Both deployed without incident. Glamsterdam is on a similar trajectory. The question is not whether the upgrade will ship, but whether the twice-yearly cadence can accommodate the scope of changes the network needs.