For people relying on an Ethereum application to access their assets, the immediate concern is whether a network upgrade could change assumptions built into its smart contracts. As reported on November 18, 2026, developers were still debating that risk; the proposed changes had not been approved.
Two proposals for the Glamsterdam upgrade, EIP-8037 and EIP-8038, aim to raise Ethereum’s per-block gas capacity from 60 million to as much as 200 million. Developers said the changes could lower costs for most users while limiting state growth. But some contracts rely on protocol assumptions that the proposals would alter.
ChainSecurity and Dedaub presented findings from AI-assisted audits that identified roughly 200 distinct classes of affected contracts, most associated with gas-cost increases rather than the proposals’ new mechanics. Ethereum Foundation developer Marius van der Wijden put the highest estimate at 10,000 to 15,000 contracts out of roughly 90 million. That is an estimate of potentially affected contracts, not a count of broken applications or confirmed losses. He loosely estimated that the value affected among contracts not yet committed to upgrading was in the low hundreds of thousands of dollars.
Application developers disputed the trade-off. Juicebox developer Jango warned that changing these assumptions could undermine composability and confidence in immutable protocols. A Uniswap developer urged reconsideration of EIP-8037. The Ethereum Foundation had already delayed Glamsterdam’s testnet fork over the gas-related security concerns.
At the time of reporting, developers said no decision had been made and expected the issue to go to an execution-layer call that Thursday. Some favored keeping the proposals, arguing that removing them could delay Glamsterdam and the broader post-quantum security roadmap by four to six months. The custody-relevant distinction is whether a contract’s access and operation depend on assumptions the network might change—not whether an estimated impact has already become a loss.
