Ethereum's Glamsterdam Upgrade Reaches Sepolia as the Mainnet Window Narrows
The largest protocol change since the Merge forks on Sepolia around 21 September, leaving client teams roughly six weeks to find what testnets always find late.
Ethereum's Glamsterdam upgrade is scheduled to fork the Sepolia testnet around 21 September, the last full rehearsal before a mainnet activation that developers have penciled in for November. It is the most substantial change to the protocol since the Merge, and the first in some years whose headline features alter how blocks are built rather than how they are paid for.
What is actually changing
Two items carry the release. The first is enshrined proposer-builder separation, which takes an arrangement that has been held together by off-chain relays and social trust since the Merge and writes it into the protocol itself. For three years the separation of the party that proposes a block from the party that assembles its contents has been a convention enforced by software nobody is obliged to run. Enshrining it removes a set of trusted intermediaries that the network has been quietly depending on.
The second is block-level access lists, which require a block to declare up front which pieces of state it will touch. That sounds like bookkeeping and is closer to plumbing: it lets clients parallelise execution and validate in stages, and it is a precondition for several of the scaling changes queued behind it.
The part worth watching
Testnet forks rarely fail in the way anybody expects. Sepolia's job over the next weeks is not to prove that Glamsterdam works — local devnets have already done that — but to surface the disagreements between client implementations that only appear under real traffic and real validator diversity. Those disagreements are routine. They are also the reason mainnet dates move.
Anyone running infrastructure should treat November as a target rather than a commitment. The changes touch block construction directly, which means block builders, relays and the MEV supply chain all have work to do that is not finished when the clients ship. Staking providers in particular have an upgrade path to test that does not appear in the protocol specification at all.
What it means for the rest of the stack
For applications, Glamsterdam should be close to invisible. For anyone operating a validator, a builder or a rollup, it is the largest coordination exercise in some time. The lesson of every previous upgrade holds: the protocol change lands on schedule far more often than the ecosystem around it does.