# Ethereum's Glamsterdam upgrade goes live on Sepolia with a 200 million gas ceiling that early blocks leave half empty

The test network rehearsal brings in enshrined proposer-builder separation and block-level access lists ahead of a mainnet activation that developers target for December, but sampled blocks used less than half of the new processing budget.

- Published: 2026-10-08T05:50:08.479Z
- Canonical: https://polylog.news/crypto/2026-10-08/ethereum-s-glamsterdam-upgrade-goes-live-on-sepolia-with-a-2
- Publisher: Polylog (Crypto desk)
- Section: crypto
- Sources: [CoinDesk](https://www.coindesk.com/tech/2026/10/08/ethereum-s-glamsterdam-test-runs-near-200-million-gas-per-block-after-upgrade), [Ethereum Foundation Blog](https://blog.ethereum.org/2026/09/17/glamsterdam-testnet-announcement), [ETH Daily](https://ethdaily.io/glamsterdam-activates-on-sepolia-testnet), [Crypto Briefing](https://cryptobriefing.com/ethereum-sepolia-forks-gloas-200m-gas/)

Glamsterdam, Ethereum's next hard fork, activated on the Sepolia test network at 13:53:36 UTC on October 6 and [finalized at 14:24 UTC](https://ethdaily.io/glamsterdam-activates-on-sepolia-testnet). Sepolia is the second of three public test networks to run the upgrade, and developers are targeting a mainnet activation in December. Within two days, the rehearsal had [raised the per-block processing budget more than threefold](https://www.coindesk.com/tech/2026/10/08/ethereum-s-glamsterdam-test-runs-near-200-million-gas-per-block-after-upgrade), to close to 200 million gas. Gas is Ethereum's unit for metering computation. CoinDesk found that the blocks it sampled used less than half of that allowance.

The fork bundles several changes that the [Ethereum Foundation lists](https://blog.ethereum.org/2026/09/17/glamsterdam-testnet-announcement) as the next step of its base-layer scaling roadmap. Enshrined proposer-builder separation (ePBS) moves the split between the validator who proposes a block and the specialist who builds it into the protocol itself. Today that split runs through trusted relays outside the protocol. Ethereum Improvement Proposal (EIP) 7928 adds block-level access lists, which record every account and storage slot a block touches along with the resulting state changes. Clients can then read from disk, validate transactions and compute state roots in parallel instead of one transaction at a time. The fork also reprices gas so fees better reflect execution costs and the cost of state growth.

On Sepolia the higher limit is opt-in. Validators signal the gas limit they will accept, which [lets developers watch](https://cryptobriefing.com/ethereum-sepolia-forks-gloas-200m-gas/) how nodes handle heavier blocks before anyone commits mainnet to the same ceiling. The gap CoinDesk measured between capacity and use is the main data point so far. On a test network, demand is synthetic, so the half-empty blocks mainly show that the network can carry the larger allowance without trouble. They do not show how much demand mainnet would actually bring to it.

The fork matters well beyond Ethereum's own fees. Ethereum holds $52.42 billion of the $93.57 billion in total decentralized finance (DeFi) value locked, according to DeFiLlama. Its largest rollups, Base and Arbitrum One, secure $15.74 billion and $11.12 billion. L2Beat shows both down over the past week, by 1.7% and 3.7%. A base layer with three times the capacity competes directly with those networks for the same transactions.

## What this means

A higher gas ceiling paired with flat demand lowers the price of each unit of block space. That cuts the amount of Ether destroyed through fees and narrows the cost advantage that general-purpose rollups sell. Rollup tokens and sequencer revenue lose the most if mainnet demand does not grow alongside capacity. Validators with modest hardware also bear the cost of processing heavier blocks, a decentralization question as much as a fee question. The market has priced Glamsterdam as a scaling win for Ether. The CoinDesk sample shows that capacity is not the constraint today, a point the consensus view tends to skip. The precedent is 2025, when validators raised the mainnet limit in steps from 30 million to 60 million instead of in one jump. The likelier outcome is that Glamsterdam reaches mainnet on schedule within months and validators then raise the limit toward 200 million in stages over the following quarters. That depends on the third test network and the first mainnet weeks showing no rise in missed slots or block propagation delays. A quick, coordinated validator move to the full 200 million at activation would prove this read wrong.

## What to watch

- Missed-slot and block-propagation data from Sepolia over the coming weeks. Stress from larger blocks would show up there first and would likely slow the mainnet schedule.
- The date of the third public test network fork. Any slip in it pushes back the December mainnet target.
- Weekly changes in the value secured on Base and Arbitrum One. Sustained declines alongside base-layer scaling would show activity starting to move back to Ethereum itself.