Ethereum (ETH) Developers Target Oct. 6, 13:53 UTC for Glamsterdam on Sepolia

Sepolia Fork Slated for Oct. 6

Ethereum (ETH) developers have provisionally set Oct. 6, 2026, at 13:53 UTC as the activation slot for the Glamsterdam upgrade on the Sepolia testnet — a date that remains explicitly conditional, because the fork has yet to run stably on any private development network. The schedule was agreed at the 186th All Core Developers Consensus (ACDC) call on Sept. 3, where participants settled on epoch 351232 as the proposed fork boundary. That epoch corresponds to the early-afternoon UTC timestamp, handing client teams, staking providers and infrastructure operators a shared planning target for the next network-wide hard fork of the Ethereum protocol.

Meeting notes from the call and follow-up reporting from protocol researcher Christine D. Kim make one point unambiguous: choosing an epoch is a coordination step, not a commitment. When developers picked the Sepolia slot, Glamsterdam had never completed a stable activation on a private devnet. The most recent attempt, Devnet-9, launched Sept. 1 with roughly 1,000 validator nodes — the largest Glamsterdam devnet by validator count to that point — and still failed to finalize, leaving blocks without the supermajority attestation that finality demands. Attention has since shifted to Glamsterdam-Devnet-11, expected to launch Monday, Sept. 14, one iteration beyond earlier plans that had centered on Devnet-10.

Glamsterdam is the next scheduled upgrade after Fusaka, which reached mainnet last December as part of the network's semiannual release cadence — the same cadence that carried the chain from the Ethereum 2.0 transition to today's rapid shipping tempo. Selecting an epoch now lets operators sequence releases, but developers can postpone the fork if the next testing phase uncovers a serious fault or client teams cannot ship reliable code. No dates exist for the Hoodi testnet or mainnet; a December mainnet activation has been discussed, yet the call's notes treat it as a possibility, not a scheduled outcome.

Devnet Failures Shaped the Conditional Date

The conditional framing traces directly to what the private testnets have exposed. Stefan Starflinger, developer operations engineer at the Ethereum Foundation, told the ACDC call that Devnet-8 surfaced a consensus-layer defect in which a block repeating its parent's hash could, in his words, “get the whole network to stop.” Devnet-9 then ran into non-finality at a scale of about 1,000 validators, forcing engineers to probe edge cases across a larger validator set than any prior Glamsterdam trial and to determine whether the cause sat in client software, validator participation or network configuration.

The execution layer carried its own baggage. Ethereum Foundation researcher Maria Silva reported an implementation problem touching EIP-8037, the proposal that reworks how the network charges gas fees for creating new state — new accounts, contracts and storage entries. The bug obliged every execution client to revise its implementation and triggered additional specification work. Client diversity compounds the workload: several independently developed execution and consensus clients must interoperate, and a fault confined to one implementation can still stall a testnet when the affected validators carry enough weight.

The ACDC #186 agenda records requests from Lido and Optimism for at least one stable day of devnet operation before any fork, listing client fixes and successful cross-client interoperability as prerequisites for Sepolia. Paritosh Jayanti, who chaired the call, cautioned that if the next devnet fails, the following week's developer calls would weigh skipping the Oct. 6 fork altogether. Teku developer Enrico del Fante backed holding off on a Hoodi date until Sepolia is settled, citing the Devnet-9 disruption. Layer-2 teams and staking operators, meanwhile, need working client releases before they can safely test their own systems against the proposed fork. Readers tracking the market in real time can follow live spot and futures prices on Gate.

December Mainnet Window Hangs on Sept. 14

COINOTAG's read is that the official paper trail now matters more than the calendar. EIP-8037's published specification describes a multidimensional gas model that separates state-creation costs from ordinary execution costs, designed to contain state growth as the block gas limit rises — and it remains under peer review, which is precisely why implementation-level bugs surfaced only at devnet scale. That makes Sept. 14 the operative gate: a clean Devnet-11 run preserves Oct. 6 and keeps the thin December mainnet path alive, while a repeat of Devnet-9's finality failure would send the fork back to the core-developer calls. With Ethereum daily transactions topping 2 million, shipping a broken fork has become costlier than delaying one.