Ethereum's Glamsterdam Upgrade May Triple Throughput But Risk Breaking Millions of Smart Contracts

Ethereum's candidate Glamsterdam upgrade aims to support roughly three times more base-layer throughput. The upgrade could silently break millions of existing smart contracts. The Ethereum Foundation said the plan would align gas charges with the resources that each operation consumes. EIP-8037 and EIP-8038 remain under formal review. Ethereum's official roadmap plans the upgrade for the fourth quarter of 2026. Fixed Sepolia, Hoodi, and mainnet fork dates have not been announced. EIP-8037 estimates that Geth's state database was about 390 GiB in January 2026. Average daily state growth rose from roughly 105 MiB to 326 MiB after Ethereum's gas limit increased from 30 million to 60 million. EIP-8037 extrapolates roughly 387 GiB of annual growth under a 200 million gas-limit scenario. The proposal says that figure could cross a cited 650 GiB performance threshold within a year from the 390 GiB starting point. The proposal describes the extrapolation as a motivating estimate rather than a measured forecast. EIP-8037 proposes a common cost of 1,530 gas for every byte of new state. It also proposes a separate state-gas dimension. A reference block limit of 150 million targets average annual state growth of 120 GiB. The proposal lists 160 GiB of annual growth as the worst case at a 200 million limit after repricing. Users would pay for execution gas and state gas. EIP-8038 would raise selected account and storage costs for access to and writes of existing state. The proposal uses client benchmarks based on a state snapshot comparable to mainnet in March 2026. The proposals use roughly threefold throughput as an engineering support target. Activation would create headroom for higher limits rather than guarantee an immediate tripling of mainnet capacity. A public dashboard covers 929,731,274 transaction replays across 4 million blocks from Dec. 3, 2024, through June 15, 2026. Researchers replayed each transaction against its canonical pre-transaction state under the existing schedule and one candidate schedule. Researchers first used each transaction's original gas limit. Researchers then tested a ceiling of up to 10 times that limit. The comparison covered success, gas consumption, logs, output, and execution traces. Under EIP-8037, 174,473,898 replays failed at the original limit but succeeded with more gas. Another 2,687,652 replays entered the potentially broken group. Under EIP-8038, 84,708,228 replays were fixable with a higher limit. Another 3,036,537 replays were potentially broken. The figures do not represent millions of separate contracts because repeated activity from one busy application can dominate the count. The dashboard defines potentially broken transactions as baseline-successful transactions that still failed under the candidate schedule at the tested 10-times ceiling. The group includes out-of-gas cases. It also includes transactions that reverted for other reasons after the new costs changed execution behavior. The classification represents counterfactual risk because users, wallets, contracts, builders, and fee markets can adapt before mainnet. The larger fixable group would require frontends, bundlers, and infrastructure providers to submit limits that reflect the new schedule. The harder cases include fixed 2,300-gas stipends. They also include hardcoded gas forwarded to internal calls. They include logic that branches on presigned transactions with fixed limits. A public outreach report identified versions of the eth-infinitism ERC-4337 EntryPoint and related smart-account stacks, including ZeroDev and Alchemy. The report also identified recurring counterfactual failures involving Across, Socket/Bungee, CoW Protocol, and 0x. Immutable contracts can require a new EntryPoint, account, factory, or validator implementation. Users may then need to migrate. Other systems may be repaired through routing, batching, or gas-accounting changes. ERC-4337 bundlers need to distinguish fixable failures from failures caused by validation guards or internal gas assumptions. Wallets, RPC providers, indexers, node tooling, and gas estimators need to incorporate the new rules. Frontends and bundlers need to stop relying on cached constants or limits calibrated to the old schedule. Both EIPs require related logic to account for the new costs. A Plataberget testnet announcement warned that tools built around a hardcoded maximum gas limit or a single gas dimension would break. A plain ETH transfer to an existing account can still use 21,000 gas. Sending value to an account that does not yet exist adds a state-creation charge at runtime. Developers need to resimulate transactions, review fixed stipends and internal call limits, and test affected paths against the candidate schedule. A published sequence scheduled the Glamsterdam fork for Aug. 20 and reported it live with the repricing schedule by Aug. 24. The sequence moves from stable devnets to the Sepolia and Hoodi public testnets and then to mainnet. The 2026 roadmap target remains subject to testing.