Proposer-builder separation - PBS - is one of those protocol changes that sounds like plumbing until you realize it determines who profits from every Ethereum transaction.
The basic mechanic: under PBS, the job of proposing a block is split from the job of building one. Validators (proposers) no longer construct the blocks they submit. Instead, specialized block builders assemble transaction bundles optimized for maximum extractable value, then auction those blocks to validators via relays. The validator picks the highest bid, signs it, and collects the payment without needing to know - or care - what’s inside.
This wasn’t a spontaneous design choice. It was a direct response to MEV. Before MEV-Boost (the off-protocol implementation of PBS that Ethereum has used since the Merge), sophisticated actors were front-running and sandwiching transactions directly at the validator level. Spreading that work to specialized builders was supposed to democratize MEV revenue - small validators could earn roughly the same block rewards as large ones, because they’re all selling to the same competitive builder market.
That part worked. Solo validators do earn more MEV revenue than they would have under the old model.
The Concentration Problem Nobody Fixed
What PBS didn’t solve - and arguably accelerated - is builder concentration. Data tracked by platforms like mevboost.pics has consistently shown that a small number of builders produce the majority of winning blocks. At various points in 2025, the top three builders by block share represented over 80% of the MEV-Boost market. The proposers are diversified; the builders are not.
This matters because builders, not validators, decide transaction ordering. If a handful of builders collude or get acquired by the same entity, they can selectively include or exclude transactions - not just optimize for profit, but enforce policy. A builder controlled by a regulated exchange might filter OFAC-sanctioned addresses. One controlled by a trading firm might favor its own order flow.
Ethereum’s roadmap includes enshrined PBS (ePBS), which would bring this mechanic into the protocol itself rather than relying on MEV-Boost’s off-chain relay infrastructure. Attester-Proposer Separation (APS) is a related proposal that would go further, separating attestation duties as well. Neither has shipped.
Why This Sits Differently Than Other Centralization Risks
Most centralization risks in Ethereum - staking pool dominance, client monoculture - operate at layers users can observe and pressure. Builder centralization is invisible to end users and operates faster than governance can respond. By the time a problematic builder achieves dominant market share, the transactions they’ve shaped are already finalized.
ePBS would at least make the rules explicit and enforceable on-chain. Until it ships, the block construction layer remains the part of Ethereum’s stack most disconnected from its decentralization claims.