The Ethereum Foundation published its Glamsterdam testnet announcement on September 28, 2026, setting October 6 at 13:53:36 UTC as the scheduled Sepolia activation time. The announcement covers a testing network, not a completed Ethereum mainnet upgrade. As reviewed on October 8, it still listed Hoodi and mainnet dates as undecided.
The central development is a coordinated test of changes that reach beyond a simple increase in transaction capacity. Glamsterdam changes how blocks are assembled, how computers check their contents and how the network charges for creating persistent data. Those pieces address different constraints, so their effects should be assessed together.
Block builders get a role inside the protocol
EIP-7732 specifies enshrined proposer-builder separation. In ordinary language, it puts the relationship between the validator proposing a block and the specialist constructing its transaction payload into Ethereum’s own rules. The builder commits to a payload and payment, then reveals the payload.
The specification also separates the timing of consensus checks from execution checks. A validator committee reports whether the promised payload arrived on time. Other validation work can happen over a longer part of the available interval. This is intended to reduce reliance on trusted middleware and ease pressure on the most time-sensitive steps.
The design introduces its own questions. EIP-7732 discusses builders withholding payloads when that behavior becomes economically attractive. Bringing an activity inside the protocol does not make every participant cooperative; the network still needs rules for late delivery, missing data and conflicting incentives.
Access lists make the work easier to divide
A second specification, EIP-7928, introduces mandatory block-level access lists. These identify the accounts and storage locations touched by a block and record the relevant changes. Knowing which data will be needed allows clients to fetch it in parallel instead of discovering every dependency during sequential execution.
The specification describes benefits for parallel transaction validation and for calculating the state root, the cryptographic commitment to the resulting ledger state. It also supports reconstructing state from the recorded changes. These are changes to how nodes organize work, rather than a promise that every application suddenly becomes faster.
That distinction matters when comparing networks. The Cardano profile and Algorand profile explain other approaches to processing and validating transactions. A headline throughput number cannot capture all the differences in workloads, hardware assumptions and security models.
Scaling also changes the price of storing data
EIP-8037 tackles the cost of adding to Ethereum’s persistent state. It proposes separate accounting for execution and state creation, with a more consistent charge for the data a transaction leaves behind. The purpose is to keep expanding storage demands from becoming a bottleneck as block capacity rises.
Under the proposal, the sender still pays for both kinds of work. Separating the counters changes how resource use is measured and how the block limit is applied. It does not mean persistent storage becomes free, or that every contract interaction becomes cheaper.
A useful consequence follows: scaling and repricing can occur together. A network can accommodate more work overall while increasing the relative cost of operations that create lasting burdens. Developers need to examine the operations their contracts actually perform, rather than assuming an upgrade has one uniform effect on fees.
What the announcement means for users and developers
The Foundation’s instructions call for Sepolia operators to run compatible execution and consensus clients. Application developers are told to test assumptions about gas accounting and contract behavior. Ordinary mainnet ETH holders were given no action to take as a result of this testnet announcement.
The remaining distinction is between a published schedule and verified results. The scheduled Sepolia date has passed, but the announcement alone does not establish that activation completed successfully or describe subsequent performance. This article therefore treats September 28 as the news event and does not infer a successful rollout from the calendar.
Readers following network fundamentals can compare the Bitcoin network’s design with the changes being tested here. The relevant next evidence is operational: compatible software, test results and a separately confirmed mainnet timetable. None of those can be replaced by a prediction about ETH’s market price.