Proposer-builder separation was introduced to Ethereum’s roadmap as a fairness mechanism. The idea was straightforward: validators shouldn’t have to compete with sophisticated actors extracting MEV from the same blocks they’re producing. Split the roles - let specialized builders construct blocks, let validators simply propose them - and you remove the pressure on solo stakers to run complex MEV extraction software just to stay competitive.
What’s emerged in practice is a builder market so concentrated it raises questions about whether PBS solved the problem or relocated it.
What the Builder Market Actually Looks Like
On any given day, a small handful of builders account for the overwhelming majority of Ethereum blocks. Data from platforms tracking MEV-Boost relay activity has consistently shown that the top three or four builders routinely construct 70–90% of blocks. These builders aren’t running on superior consensus logic - they’re winning because they have privileged relationships with order flow. Wallet providers, DEXs, and other dApps route transactions directly to preferred builders through private order flow agreements, cutting out the public mempool entirely.
This matters because validators running MEV-Boost - which is to say, most validators - are now dependent on this small cohort of builders to deliver competitive bids. A validator who rejects all builder bids and builds their own blocks locally sacrifices significant revenue. In practice, that means the block contents of Ethereum are being determined not by the validator set, but by a few private entities with preferential transaction access.
Inclusion Lists Were Supposed to Help
EIP-7547 proposed inclusion lists as a partial fix: validators could specify transactions that builders must include in any block they submit. The intent was to preserve validator agency over censorship resistance without requiring validators to do full block construction themselves.
The implementation challenges, though, are non-trivial. Inclusion lists add latency to an already tight slot window, and there’s active debate about whether the mechanism can be gamed by builders who technically satisfy list requirements while still manipulating transaction ordering around them.
The Underlying Tension
PBS assumes that separating proposal from construction is enough to distribute power. But construction is where the value is extracted and where order flow relationships get monetized. The separation succeeded in insulating validators from having to run MEV bots - but it handed block construction to entities that are, in structural terms, less accountable than validators.
Validators are staked, slashable, and publicly identifiable on-chain. Builders are not. That asymmetry isn’t incidental to the current centralization - it’s probably the cause of it.