Glamsterdam Upgrade: What Changes for Ethereum Security

By Owen Thurm Published November 6, 2026 6 min read

Quick answer. Glamsterdam would add enshrined proposer-builder separation, block-level access lists and higher state gas costs. Creating storage, accounts or contracts would cost 5 to 8 times more. It runs on Sepolia; mainnet has no date, pending a gas review.

Watch: Glamsterdam Upgrade: What Changes for Ethereum Security 1:45

Video transcript

Glamsterdam would make creating storage, accounts or contracts on Ethereum five to eight times more expensive. Unchanged contracts could behave differently.

Glamsterdam runs on Sepolia. The Hoodi fork was postponed, and mainnet has no date while developers study the gas changes.

ePBS and block-level access lists are the headline features. For contract security, the proposed gas repricing matters more.

These proposed numbers may still move, but writing a new storage slot would rise from about twenty-two thousand gas to about one hundred ten thousand. Creating accounts and contracts jumps further.

ChainSecurity found projects that would be vulnerable. Their warning: a failure path that used to revert can now complete.

The Ethereum Foundation lists eight broken invariants. Most findings involve fixed gas budgets and the transfer stipend, but failure handlers and gasleft billing matter too.

ePBS replaces trusted middleware with staked builders who pay proposers through the protocol. Withholding and reorg risks remain, and stakers gain a new duty.

Offchain tools can break too. Watch the new ETH transfer logs, re-estimate gas instead of caching it, and check outcomes, not inclusion.

Search for hard-coded gas, re-estimate state-creating paths, replace transfer and send with checked calls, and run the Foundation's invariant checks.

Guardian's USDT0 review found an unwrap that could leave funds parked after running out of gas. Read the full article, or talk to Guardian before the fork.

Glamsterdam is Ethereum's next network upgrade after Fusaka. The name joins Amsterdam, the execution layer upgrade, and Gloas, the consensus layer upgrade.

On November 6, the Ethereum Foundation said core developers had delayed the fork process. ChainSecurity had found a few projects that would be vulnerable after the upgrade.

Here is what is decided, what is still open, and what to check in contracts, wallets and monitoring. Facts are as of November 6, 2026.

Status and Timing

Network	Glamsterdam status
Sepolia testnet	Activated October 6, 2026
Hoodi testnet	Postponed October 22; it was set for October 26
Mainnet	No date set

Hoodi was postponed for "further investigation of interactions of EIP-8037 with mainnet smart contracts." A new schedule will be announced later.

ethereum.org's Glamsterdam page, last updated October 16, says mainnet is expected in Q4 2026 but not confirmed. That predates the postponement.

The scope sits in EIP-7773, the upgrade's meta EIP, which is in Review. It schedules 18 EIPs: 17 were in Last Call on November 6, and EIP-7688 in Review.

Scheduled is not frozen. The Ethereum Foundation says it may slightly change EIP-8037 and EIP-8038, depending on what auditors and teams report.

What Is Scheduled

These are the 18 scheduled EIPs as specified today. ethereum.org notes the scope can still change before mainnet.

Area	EIPs	What would change
Block building	7732, 8282	Builders become staked protocol participants (ePBS)
Execution	7928	Blocks carry enforced access lists and state diffs
Gas	2780, 7778, 7976, 7981, 8037, 8038	New prices for transactions, calldata, access lists and state
EVM	7708, 7843, 7954, 7997, 8024, 8246	Transfer logs, SLOTNUM, stack opcodes, 64 KiB contracts, CREATE2 factory, SELFDESTRUCT
Validators	7688, 8045, 8061	Data structures, slashed proposers, exit and consolidation churn

ePBS and block-level access lists are the headline features. For contract security, the gas EIPs matter more.

Proposed Gas Changes Break Assumptions

EIP-8037 would raise the cost of creating state, targeting 120 GiB of state growth a year at a 150M gas limit. EIP-8038 would raise the cost of accessing and writing existing state.

Operation	Today (Osaka)	Glamsterdam (subject to change)
Write a new storage slot, cold	22,100	110,020
Change an existing slot, warm / cold	2,900 / 5,000	10,100 / 12,100
Send ETH to an empty address	25,000	183,600 plus execution
CREATE to an empty address	32,000	195,600
Code deposit, per byte	200	1,530

These are the Ethereum Foundation's measurements on a pinned spec, and they change if the spec changes.

A plain ETH transfer to an existing account stays at 21,000 gas under EIP-2780.

EIP-8037 would also add a reservoir. When a transaction's gas limit exceeds 16,777,216, state gas is kept apart from gasleft().

ChainSecurity's finding follows from this: "Even with unchanged contract code, a failure path that previously reverted can now complete."

The Ethereum Foundation lists eight invariants these changes would break. Most findings so far involve the first two below:

Fixed gas budgets: hard-coded limits in contracts, settings, signed messages, bots and relayers can run short.
Stipends: the 2,300 gas stipend of transfer and send cannot create a storage slot.
Failure handlers: after a subcall runs out of gas, the caller may still afford to record a permanent failure.
Gas refills: clearing a fresh slot, or a reverting call, hands gas back at once.
gasleft() billing: gas left can rise during a call, so subtracting readings misprices work.
7702 replays: a delegation that runs out of gas is rolled back, and its signed authorization can be reused.

For how storage and gas work underneath, see how the EVM works. For delegation risks, see EIP-7702 vs ERC-4337.

ePBS and Builder Trust

Most proposers already outsource block building through trusted middleware, EIP-7732 notes. ePBS moves that exchange into the protocol.

Builders become staked entities, from as little as 1 ETH. Each commits to a payload and a payment, and the in-protocol payment comes out of its beacon chain balance.

A Payload Timeliness Committee of validators attests whether the builder revealed its payload and blob data on time. ethereum.org says this widens the propagation window from about 2 seconds to about 9.

The EIP itself names risks that remain:

Withholding: a builder may withhold a payload when that pays, which can cause missed slots.
Reorgs: proposers and attesters with consecutive blocks and over 20% of stake could reorg a builder's payload.

Stakers take on a new duty, payload timeliness attestations, and need updated execution and consensus clients. EIP-8045 stops slashed validators from proposing; EIP-8061 raises exit churn.

Wallets, Indexers and Monitoring

The Ethereum Foundation warns that offchain components can break too, for example through changes to receipts.

Transfer logs: EIP-7708 makes every ETH transfer emit an ERC-20-style Transfer log from 0xfffffffffffffffffffffffffffffffffffffffe. Do not index that address as a token.
Estimates: a 7702 authorization for an empty account would rise from 25,000 gas to 235,606. Re-estimate rather than cache.
Included is not executed: under EIP-2780, a transaction that cannot cover its runtime charges is still included and charged, then reverted. Check outcomes, not inclusion.
Contract size: the limit would rise from 24 KiB to 64 KiB. Larger code will not deploy on chains without the change.
SELFDESTRUCT: EIP-8246 would end its last ways of burning ETH, preserving the balance.

Block-level access lists record every account and slot a block touches, with post-transaction values. Monitoring tools gain a new data source.

Readiness Checklist

Search contracts, configs, signed payloads and bots for hard-coded gas.
Re-estimate every path that creates storage, accounts or contracts.
Replace transfer and send to contracts with checked calls.
Review each failure path that marks work as permanently done.
Keep reentrancy locks in transient storage, or at non-zero resting values.
Stop charging users by gasleft() differences.
Tie 7702 setup to the user's signature; check account code afterwards.
Update indexers and alerts for transfer logs and receipts.
Run the Ethereum Foundation's invariant checks, and report breakage privately as it asks.

Make gas limits adjustable where you can. The numbers may move again before mainnet.

How Guardian Helps

Guardian's Managed Security matches security advisories to the stack you run, assesses exposure and gets the patch through with your team.

Helix, Guardian's AI auditor, runs continuously across contracts, backend and infrastructure. Only human-verified findings reach you, each with a fix PR.

For a one-off check before the fork, Guardian's smart contract audit covers the contracts, and its upgrade reviews cover migrations and parameter changes.

Guardian's public USDT0 Glamsterdam gas price review (March 2026, on the spec at the time) shows the work. It found that an unwrap running out of gas could leave USDT0 parked until a retry.

Frequently Asked Questions

When is the Glamsterdam upgrade?

There is no mainnet date. Glamsterdam activated on Sepolia on October 6, 2026. The Hoodi fork set for October 26 was postponed, and on November 6 the Ethereum Foundation said the fork process was delayed to study the gas changes.

What is ePBS?

Enshrined proposer-builder separation, EIP-7732. Builders become staked protocol participants who commit to a block and pay the proposer through the protocol, replacing trusted middleware. A committee of validators attests whether the builder revealed the block on time.

Will Glamsterdam break my smart contracts?

Possibly, if they rely on fixed gas amounts, the 2,300 gas stipend, gasleft() arithmetic or failure handlers sized for today's prices. The Ethereum Foundation lists eight invariants that would break. Check them before mainnet; the numbers may still move.

Do I need to do anything with my ETH?

No. ethereum.org says there is no need to convert or upgrade ETH, and warns that anyone telling you to is trying to scam you. Node operators and validators will need updated clients.

Sources

Get your contracts ready for Glamsterdam

Guardian matches security advisories to the stack you run, assesses your exposure and works the fix through with your team.

Talk to Guardian Explore Managed Security