In March 2025, the Ethereum Foundation published its quarterly budget report. Buried on page 17, a footnote revealed that direct protocol development grants had been cut by 28% year-over-year. The Foundation’s own spending on core R&D had shrunk to its lowest share since the 2020 merge preparations. No official announcement accompanied the change. No EIP was filed. Yet for those reading the ledger closely, this was the loudest signal yet that a quiet tectonic shift is underway—power is slowly migrating away from the Foundation, coalescing around a new set of nodes: client teams, staking pools, infrastructure providers, and a handful of influential L2 protocols.
To understand why this matters, we need to step back. For nearly a decade, Ethereum’s governance has been a paradox. The network prides itself on being “decentralized,” yet its most critical decisions—which EIPs get prioritized, how client teams coordinate, where treasury funds flow—have historically radiated from a single point: the Ethereum Foundation headquartered in Zug. This was never purely a control structure; it was a narrative convenience. The Foundation acted as the benevolent steward, the “neutral arbiter” that kept the chaos of competing interests in check. But as the network matured through The Merge, the Shanghai upgrade, and now the pectra era, the Foundation’s gravitational pull has weakened. The incoming post-merge world distributes power not through hierarchical decree, but through the brute force of market dominance and technical dependency.
The first node to emerge is the client oligarchy. Geth alone now commands over 70% of execution layer clients. Nethermind, Besu, and Erigon hold the remainder. While diversity is preached, reality is stark: if Geth’s lead developers decide to delay or push a controversial EIP, they hold an de facto veto. I observed this dynamic firsthand during my audit work in early 2023, when a minor client disagreement over gas limit adjustments stalled a proposed improvement for three months. The Foundation mediated, but the real leverage belonged to the client team controlling the majority of the node count.
The second node is the staking liquidity cartel. Lido’s stETH now accounts for roughly 33% of all staked ETH. Combined with Rocket Pool, Coinbase, and Binance, the top five staking pools control over 60% of the consensus layer’s voting power. In a proof-of-stake system, staking power is governance power. The Foundation’s ability to steer decisions on slashing conditions, validator set dynamics, or withdrawal queues is increasingly circumscribed by the interests of these large stakers. They don’t need to vote on EIPs; their mere existence shapes the incentive landscape.
The third node is infrastructure. Infura and Alchemy serve as the default RPC gateways for the vast majority of dApps and wallets. If these infrastructure providers blacklisted a certain type of transaction or upgraded to a new client version ahead of schedule, they could effectively fork the network’s user experience without changing a line of code. The Foundation cannot ignore them.
Together, these nodes form a “multi-node governance” that is neither formalized nor transparent. It operates through what I call “structural consensus”—a set of unwritten rules enforced by market share and technical necessity. The Foundation’s budget cuts are not a retreat; they are an acknowledgment that its role has evolved from commander to concierge. It now funds ecosystem grants, educational outreach, and academic research, while the real power over protocol direction has migrated to those who run the network day-to-day.
The contrarian angle: This is not necessarily a healthy evolution. Multi-node governance may reduce the single point of failure that the Foundation represented, but it introduces a new set of dependencies that are equally fragile. The three nodes—client dominance, staking concentration, infrastructure gatekeeping—are themselves highly centralized. Geth’s 70% market share is a single point of failure. Lido’s 33% staking share is a systemic risk that could freeze withdrawals or trigger a cascading slashing event if misconfigured. Infura and Alchemy are centralized corporate entities subject to legal pressure in their jurisdictions.
Liquidity flows, but trust evaporates. The narrative that “decentralization increases with multi-node governance” is, at best, incomplete. What we are witnessing is a shift from one form of centralization (the Foundation) to a triopoly of centralization (clients, stakers, infrastructure). The Foundation’s waning power is a story of erosion, not liberation. The real question is whether this new configuration will be more resilient. History suggests that when control coalesces in a few hands, the network becomes vulnerable to capture—not by a benevolent steward, but by profit-maximizing entities who have no loyalty to the broader mission.
So what should we watch for? The first real test will come with the next contentious EIP—perhaps EIP-7781 that aims to reduce block times, or a proposal to adjust the issuance curve. If the Foundation attempts to push one direction and the client oligarchy pulls another, we will see which node truly governs. A fragmented EIP process would signal that governance has fractured, not diversified. For now, the silence from Zug is deafening.
Don’t trade the chart; trade the story. The story of Ethereum’s governance is shifting from “one foundation, many voices” to “no foundation, few nodes.” That narrative, if it solidifies, will redefine how institutions and regulators perceive Ethereum’s maturity—and ultimately, its risk profile.
In the end, Code is law, but narrative is truth. And the narrative of Ethereum’s silent fracturing is one that demands more attention, not less.
Based on my experience auditing Curve’s liquidity pools in 2020, I learned that the most dangerous assumptions are those that go unchallenged. The assumption that the Foundation was the sole seat of power was always a convenient fiction. Now that the fiction is fading, we must be honest about what replaces it. The new governance model is not a utopia; it is a messy, realpolitik negotiation between client teams, stakers, and infrastructure providers. Whether that is better than a single foundation remains an open question—one that will be answered not in a press release, but in the quiet moments when an EIP is delayed, a client update is rejected, or a staking pool gains an extra percent.