Ethereum’s Glamsterdam upgrade has been pushed to 2027, revealing a widening gap between the speed of Ethereum development and the network’s ability to deliver major upgrades on schedule.
On Nov. 20, Ethereum Foundation researcher Toni Wahrstätter said the upgrade, which the foundation had initially targeted for the first half of 2026, will not reach mainnet before mid-January 2027 after testing delays disrupted its timeline.
In a detailed assessment of the setback, Wahrstätter acknowledged that Ethereum developers had repeatedly underestimated how long significant network upgrades would take, even as productivity in the ecosystem reached record levels.
He said:
“Nobody did a bad job, but I think we fooled ourselves, twice.”
His findings point to a growing tension between faster development and the ability to oversee increasingly complex upgrades.
Glamsterdam exposes limits of faster development cycle
According to Wahrstätter, monthly contributions to Ethereum’s technical specifications have risen between twofold and 3.5-fold since 2024, reflecting a notable acceleration in development activity.
The improvement extended to testing. Glamsterdam completed 12 development networks, or devnets, in 160 days, compared with seven devnets over 285 days for Pectra, an earlier Ethereum upgrade.
However, Wahrstätter argued that the acceleration created another problem: developers could produce specifications, code and tests faster than colleagues could review them and understand their implications.
That challenge became more pronounced with Glamsterdam’s unprecedented size. The upgrade contained approximately 6,080 lines of execution-layer specifications, compared with 2,318 for Prague, the previous record.
In a separate count of proposal specifications, Wahrstätter cited ePBS and block-level access lists, or BALs, at approximately 11,900 and 9,300 lines, respectively. The other 16 proposals collectively accounted for about 3,600 lines.
The complexity was also reflected in EIP-8037, a proposal that changes how Ethereum charges for creating new blockchain state. Although its specifications contained roughly 1,700 lines, the proposal required about 30,000 lines of tests, more than any other proposal in the upgrade.
Together, these figures suggest that faster code production did not translate into faster network upgrades.
Wahrstätter’s historical analysis showed that every Ethereum upgrade since Dencun took more than 400 days from finalizing its scope to reaching mainnet. For context, the three most recent upgrades took between 433 and 475 days, even though their technical scope was smaller than Glamsterdam’s.
Ethereum set its shortest upgrade target in years for Glamsterdam
Despite that record, Glamsterdam was assigned a development target of just 320 days, significantly shorter than timelines achieved by previous upgrades.
Wahrstätter found that virtually every Ethereum upgrade with an announced target had missed its deadline. Fusaka was the exception, but only after developers removed Ethereum Object Format (EOF), a major component originally planned for inclusion.
Pectra similarly required developers to divide its original scope across separate upgrades.
Glamsterdam took a different approach. Developers kept adding proposals to successive testing environments while work on its two main components, ePBS and BALs, continued.
Unlike earlier upgrades, the team made no comparable reduction in scope to protect its development timeline.
As a result, Glamsterdam reached public testing before its specifications had stabilized, creating additional work at a stage when developers would ordinarily be preparing for deployment.
Wahrstätter’s analysis showed that Ethereum upgrades between Altair and Deneb recorded no specification changes after their first public testnet.
Pectra recorded two changes, while Fusaka recorded one. Glamsterdam registered 10 in just six weeks.
The difference also appeared in client development. Work on Pectra declined substantially after public testing began, whereas Glamsterdam-related activity increased 34% following its Sepolia activation, alongside a 25% rise in client commits.
Those late changes became consequential after security researchers identified problems involving EIP-8037 and EIP-8038, which alter how Ethereum charges for certain operations.
The proposals could disrupt assumptions embedded in existing smart contracts, prompting developers to postpone Glamsterdam’s Oct. 26 activation on the Hoodi testnet.
The setback pushed the expected interval between Glamsterdam’s first public testnet and mainnet activation beyond 100 days, compared with the 4- to 10-week window typical of previous upgrades, according to Wahrstätter.
During their Nov. 19 meeting, Ethereum developers tentatively targeted Dec. 16 or 17 for Hoodi testing, with mid-January 2027 emerging as the earliest possible mainnet date if the specification stayed unchanged.
Wahrstätter argued that preventing further delays would require developers to base release targets on historical performance and set firm deadlines for dropping unfinished features and completing specifications before public testing begins.
He also questioned treating successful devnet deployments as evidence that proposals were sufficiently mature for inclusion in an upgrade. He said:
“I’m sure we can ship faster, but for that, we must ship smaller.”
The next decision comes on Nov. 26, when Ethereum’s core developers are expected to determine whether Glamsterdam can proceed with its proposed December Hoodi activation.
A further postponement would place additional pressure on a release schedule that had already extended well beyond its original target.

