Two rival proposals filed on Sunday aim to stop Glamsterdam, Ethereum’s next upgrade, from making deployed contracts incompatible. Leo Glisic would keep the legacy gas schedule for existing contracts, while Giulio Rebuffo would charge everyone the new prices but keep old execution rules. Both drafts remain unscheduled as the Hoodi activation was pulled back.
🔑 Key takeaways
- EIP-8037 and EIP-8038 sharply raise the cost of creating state: the base charge for a contract rises from 32,000 to 183,600 gas.
- ChainSecurity flagged thirteen deployed contracts that could be affected, a floor rather than a census.
- Two competing drafts, EIP-8488 (Glisic) and a dual-counter design (Rebuffo), set no target date.
- The Hoodi activation planned for 26 October 2026 was postponed with no replacement date; mainnet remains pencilled in for Q4 2026.
A new gas schedule for permanent state
Glamsterdam combines Amsterdam, the execution-layer upgrade, and Gloas, the consensus-layer upgrade. Following Fusaka, it targets transaction parallelisation, higher network capacity and tighter control of the state database through new storage pricing. Other pillars include ePBS (enshrined proposer-builder separation, EIP-7732) and block-level access lists (EIP-7928 and EIP-8159). The contested part is the price of permanent state.
EIP-8037: charging for state creation
EIP-8037 harmonises and raises the cost of creating state (accounts, storage slots and deployed bytecode), and bills it separately from ordinary execution. The goal is to keep state growth at around 120 GiB per year. A separate reservoir splits compute gas from storage gas. Glamsterdam is meant to allow a limit of 200 million gas, with developers testing a benchmark limit of 150 million.
EIP-8038: costlier state reads and writes
EIP-8038 raises the price of reading and writing state, including for a slot that is already non-zero. The table below summarises the orders of magnitude reported by ChainSecurity.
| Element | Previous schedule | New schedule |
|---|---|---|
| Base charge for contract creation | 32,000 gas | 183,600 gas |
| Cold write to a new storage slot | 22,100 gas | 110,020 execution gas, when the reservoir cannot cover state creation |
| Gas limit target | Not specified | 200 million (test benchmark: 150 million) |
For a typical contract, the increase is not a one-off surcharge: it changes the price of every new state slot, before code storage and execution costs are added. Developers therefore need to revisit gas budgets sized under the old model.
Insufficient-gas griefing: a wrong decision recorded on-chain
Under the two-reservoir model, when a call consumes too much gas or reverts, the reservoir returns to its value at call entry: gas previously debited to create state becomes available again. Contracts written for the old single-gas model do not anticipate this. A fixed budget can become insufficient, and a contract that catches the failure may record a failure even though the call would have succeeded with more gas.
The danger is therefore not just a crash but a wrong decision written on-chain. A malicious caller can supply enough gas for the failure to be logged, but not enough to complete the operation. ChainSecurity calls this “insufficient-gas griefing”.
Four mechanisms
- Direct call: the reservoir passes through in full, while the EIP-150 rule, which caps forwarded gas at roughly 63/64, applies only to the execution balance; the GAS opcode reports only the latter. A failure handler that reverted for lack of gas can therefore validate a failure.
- Slot clearing: it can trigger an immediate refund of state-creation gas, whereas Octaki only settled that refund at the end of the transaction.
- Intermediary: a contract that propagates a REVERT can return to the caller the execution gas consumed to create state.
- Budget sizing under Osaka: the higher cost of the target can make a budget set under the old schedule insufficient. A pre-check guarantees the amount forwarded, not that it is enough.
Protection families, affected contracts and rival drafts
Thirteen contracts, five protection families
ChainSecurity flagged thirteen deployed contracts that could be affected, but stresses that the figure is a floor, not a census. The firm sorts protections into five families:
- A: a gas check before the call. It remains valid if the budget covers every path permitted by the new schedule.
- B1: a check after the call against a fixed requirement. Discouraged.
- B2: a check after the call against the initial gas. Also discouraged.
- C: trust in revert, the least protective option.
- D: propagating the failure by reverting the whole operation, preferred whenever the design allows it.
Schemes B1, B2 and C infer the cause of a failure from remaining gas, an assumption weakened by refunds and reservoirs. Contracts involving escrow, withdrawal requests, batch payments or message relaying are cited as priority cases to review.
Criticism goes beyond these examples. Security researcher pcaversaccio wrote on X that “imho Glamsterdam should not be shipped in its current form, period”, after flagging the gas increases to core developers two weeks earlier. banteg pointed to a list of 200,000 affected contracts in the ethereum/repricing-impact repository and argued that several high-profile protocols are affected. The two figures are not directly comparable: the first is a list of potentially affected contracts, the second a set of concrete failure-logic examples.
Two drafts, two logics
Leo Glisic’s EIP-8488 splits Ethereum’s instruction set in two. According to the draft, “Existing contracts keep the legacy schedule”, while “new contracts must use V2 opcodes that carry the EIP-8037/8038 gas schedule”. Existing applications would keep current prices, and new contracts would have to use V2 opcodes.
Giulio Rebuffo’s proposal charges the new schedule to everyone but lets older contracts keep running as if nothing had changed. Each transaction would carry two gas counters: one tracks execution under the legacy rules, the other bills the transaction on the new schedule. The draft says “billing uses the new schedule and determines the transaction fee”, but also that “no execution-client implementation or consensus tests have been run for this design”. It gives no target date.
Both texts remain drafts. The next private test, estimated at 16 to 20 November at the earliest, had been scheduled before the two drafts were filed.
Timeline, the Hoodi delay and market context
The sequence began with Platåberget, announced on 17 August 2026 for state-gas tests. Sepolia was due to activate on 6 October at 13:53:36 UTC, and the new rules were already live there by mid-month. On 13 October, ChainSecurity published “Glamsterdam Hard Fork: When Spent Gas Comes Back”, with EIP-8037 and EIP-8038 in Last Call. On 16 October, the Ethereum Foundation announced Hoodi with its gas limit raised to 200 million.
Hoodi was due to activate on 26 October 2026 at 17:42:48 UTC, at epoch 132,352. On 22 October, developers pushed it back to All Core Developers Execution call #248. The Last Call review ended on 1 November, although specification changes remained possible. On 3 and 4 November, the roadmap still read “Testing on devnets”, with a mainnet activation in Q4 2026 and no confirmed date.
“Core developers decided to delay the upcoming Hoodi hard fork on today’s ACDE call”
Alex Stokes, Ethereum core developer
The Foundation presents the delay as a configuration rollback rather than a chain reversal: Hoodi’s history remains intact, and the date was only removed from client configurations. The stated reason was “further investigation of interactions of EIP-8037 with mainnet smart contracts”. Unlike Glamsterdam’s earlier deposit-routing bug, a routing defect caught during testing, this issue sits in the economics of the EVM. No action is required from ETH holders or mainnet users.
On markets, the sources do not quantify any ETH price or volume reaction to the delay or to the two proposals. IG recalls a pattern seen before the Merge, Shapella and Pectra upgrades: appreciation ahead of the fork, then a 10 to 15% pullback in the days after activation, attributed to profit-taking. That statistic is not specific to Glamsterdam. The same source projected an average drop of about 78.6% in L1 fees via EIP-7904, an older bundle that also included EIP-8011 and EIP-8032, whereas the current file covers EIP-8037 and EIP-8038; the sources do not explicitly reconcile these two projections. Delta sees the direct market effect as still uncertain and recommends tracking gas costs and network throughput for the execution timing of permanent contracts and ETH options.
Timelines also diverge. The official roadmap lists Hoodi on 27 October as the next milestone, while Shattered reports that no replacement date had been set. Crypto APIs projected May to June 2026, IG late August or Q3, and Delta cites both a Q3 delay and a May to June target. These forecasts predate the Hoodi delay. The official page says scope is frozen, but the meta-EIP remains in draft and could still change before mainnet.
Outlook and scenarios
The debate pits two logics against each other: protecting existing applications through separate schedules, as EIP-8488 proposes, or unifying pricing by separating billing from execution, as the dual-counter design proposes. In the first case, complexity shifts to new contracts and client software; in the second, every transaction can become more expensive, and the design has no implementation or consensus tests yet.
Three paths remain open, none confirmed by the sources: a mainnet in Q4 2026 if the drafts are settled and tested quickly; a further delay if audits uncover more cases than the thirteen examples; or a narrower activation that leaves out the most disputed repricing. For developers, the priority is clear: audit every gas-dependent path and favour propagating failures through revert wherever the design allows.
Sources
- CoinDesk, November 9, 2026
- Crypto APIs, Glamsterdam blog
- ethereum.org, Glamsterdam roadmap
- Shattered, EIP-8037 delay
- IG, Glamsterdam investor guide
- Delta Exchange, Glamsterdam status review
Cet article est publié à titre informatif et éducatif. Il ne constitue en aucun cas un conseil en investissement. Faites vos propres recherches (DYOR) avant toute décision.
