Glamsterdam hits Plataberget on August 20, and hardcoded gas limits are the first casualty

The Ethereum Foundation launched Plataberget on August 17, 2026, a public testnet built specifically to rehearse the Glamsterdam upgrade, and it started in mainnet's current pre-fork state rather than already running the new rules ^[1][8]. Glamsterdam forks on that chain on August 20, 2026 ^[2], which gives operators three days of pre-fork chain to test against before the transition happens live ^[17].

The part that turns this into a work item rather than a curiosity is the gas repricing that ships alongside enshrined proposer-builder separation and block-level access lists ^[3]. According to the Foundation, wallets, blockchain indexers, gas estimators and anything else that bakes in a fixed maximum gas limit will stop working correctly after the upgrade and needs updating ^[4]. The Foundation pointed developers at EIP-8037, which introduces a state gas dimension, plus EIP-2780 and EIP-8038 for the decomposed costs ^[5]. That is a specification reading job for whoever owns your fee logic, and it is cheaper to do it now than after Sepolia.

The two structural changes are worth understanding because they explain why the repricing exists. EIP-7732 moves block building into the protocol itself ^[10]. Everstake, a staking provider, puts the share of Ethereum blocks currently assembled off-chain through MEV-Boost relays at about 88%, with no protocol-level fallback for validators when relays censor transactions or fail to deliver ^[11]. The replacement attests that a block arrived on schedule using on-chain builder identities, signed bids and a new payload-timeliness committee ^[12]. EIP-7928, the second change, records every account and storage slot a block touches along with the resulting state ^[13].

Together those two are the groundwork for raising gas limits from 60 million toward 200 million, roughly a 3.3x increase in block capacity ^[14][18]. Glamsterdam also raises the maximum deployed contract size from 24 KiB to 64 KiB and the initcode cap from 48 KiB to 128 KiB ^[15]. If you have been splitting contracts to fit under 24 KiB, that constraint moves by a factor of about 2.7 ^[19].

Plataberget is not a short window. The Foundation says it will be up for at least a few months ^[6], and it differs from Hoodi and Sepolia in being purpose-built for this fork and open to the public rather than restricted to private devnets ^[7]. That combination, public access plus a populated chain that has not forked yet, is the only place to observe the transition itself rather than the steady state after it.

What to watch: the community feedback and client software aggregation stage that follows, expected to take about another month, then progression to Sepolia and Hoodi, and only then mainnet ^[9]. Each of those steps is a chance for the repricing parameters to move, so teams should be building against the specs rather than against a single testnet's observed numbers. Further out, the ecosystem is already tracking the Hegota upgrade due in 2027 ^[16]. The near-term test is narrower: whether your gas estimator returns sane values on August 20.