On May 18, 2026, researchers Carl Beek and Julian Ma announced on X that they are leaving the Ethereum Foundation, extending a run of departures that has put the nonprofit's protocol stewardship capacity under fresh scrutiny. The exits are the latest in a pattern that now includes at least eight high-profile departures since February, coinciding with the foundation's publication of a 38-page organizational mandate that explicitly redefines its role in the ecosystem.
What the protocol cluster actually does
The EF's Protocol Cluster is the organizational layer that converts Ethereum Improvement Proposals from research artifacts into shipped code. Its members coordinate with client teams — Prysm, Lighthouse, Geth, Nethermind, and others — schedule devnet runs, align on specification timelines, and run the fork coordination process that is required to synchronize a network with no single operator. None of that work lives on-chain. It lives in GitHub repositories, Ethereum Magicians threads, bi-weekly All Core Devs calls, and the informal relationships that allow a global set of independent clients to ship a breaking protocol change on the same date.
Beek, who will finish his role on May 29, spent seven years at the foundation, including core work on the Beacon Chain — the consensus-layer architecture that underpins Ethereum's proof-of-stake transition. Ma, departing after roughly four years, worked on FOCIL (Fork-Choice enforced Inclusion Lists) and faster Layer 2 confirmation research. Earlier in May, Barnabé Monnot, Tim Beiko, and Alex Stokes also announced departures from the protocol cluster, according to the EF's Protocol Cluster Updates post that month. Beiko, in particular, had been the primary public coordinator of the All Core Devs process for years. Trent Van Epps, who organized Protocol Guild — the independent funding collective that channels resources to core developers — left earlier this year, as did Josh Stark, a senior management figure whose final day was April 30.
The departure that drew the most community debate was Dankrad Feist in October 2025. Feist, who led the development of Danksharding and co-authored EIP-4844, announced he was moving to full-time work at Tempo Labs — a Layer-1 blockchain for payments built by Stripe and Paradigm — while staying on as a part-time Research Advisor to the Protocol Cluster at the EF. His personal website confirms this dual arrangement. CoinTelegraph and Blockworks both covered community reaction to the move at the time.
The Mandate and what it says about structure
The EF published the 38-page EF Mandate on March 13, 2026 (blog.ethereum.org/2026/03/13/ef-mandate). The blog post describes it as "part constitution, part manifesto, and part guide for the Ethereum Foundation." Its central thesis is explicit: the EF is not Ethereum's owner, but one steward among many. The document commits the foundation to keeping Ethereum "censorship resistant, open source, private, and secure" — an acronym it renders as CROPS — and frames its job as ensuring those conditions hold, not managing outcomes.
The mandate's framing has a direct bearing on how the departures should be read. If the EF is genuinely trying to reduce its own centrality, then researchers leaving for independent organizations, client teams, or competing chains is, by that logic, consistent with the model rather than a rupture of it. But the argument has a structural limit: the transition away from EF-housed coordination works only if equivalent coordination capacity exists elsewhere. During Glamsterdam devnet — the current development cycle, pushing through L1 scaling improvements — that handoff is happening in real time with several of the people who ran the prior cycle now gone.
Tomasz Stańczak stepped down as co-Executive Director on February 13, 2026, via an EF blog post, approximately eleven months after taking the role alongside Hsiao-Wei Wang in March 2025. He had been brought in during a 2025 restructuring aimed at addressing community criticism about the foundation's execution pace and transparency.
Community reaction and the question being asked
The post that circulated most widely came from pseudonymous X commentator @DefiIgnas, who asked directly why so many people are leaving the Ethereum Foundation. The question circulated without a satisfying answer from the organization — the EF did not issue a public response to the cluster of May departures by the time of publication.
Bankless noted that community concern centers on the gap between the foundation's stated rationale — that the departures fit a planned decentralization of stewardship — and the absence of visible replacement infrastructure. The argument that the EF was always one steward among many is harder to make when the individuals who did the actual coordination work are leaving in a cluster during an active fork cycle.
Where Vitalik's attention is
The same day Beek and Ma announced their exits, Vitalik Buterin published a lengthy blog post at vitalik.eth.limo arguing that AI-assisted formal verification could become the definitive response to software security risk. His post described mathematically verified software as a practical near-term goal enabled by improved AI tooling, and specifically pointed to Ethereum infrastructure, zero-knowledge proof systems, and consensus mechanisms as priority targets. "AI gives you the ability to write large volumes of code at the cost of accuracy, and formal verification gives you back ... accuracy," he wrote.
The timing is not evidence of disengagement. Buterin has maintained a research-focused posture throughout the EF's organizational changes; the formal verification post is consistent with that pattern. But it is context for where the foundation's most public intellectual energy is pointed while the protocol coordination layer is being rebuilt.
What actually needs answering
The structural question the departures raise is not whether senior researchers leave organizations — that is normal — but whether the EF's protocol coordination function can be distributed without a gap in the handover. The Mandate's answer, at a philosophical level, is that Ethereum is robust enough to survive any single organization's contraction. The practical test is whether client teams, independent researchers, and Protocol Guild's successor structures can absorb the coordination surface that was concentrated in the EF's Protocol Cluster. That question will be answered by the Glamsterdam fork timeline and whatever comes after it, not by the organizational documents being published alongside it.