The fee you didn't know you were paying
You submit a swap on a decentralised exchange. The transaction sits in the public mempool for a few hundred milliseconds, visible to every bot on the network, and in that window something happens to it. A searcher's algorithm spots the incoming order, calculates the price impact, and wraps your trade inside two of its own: one before, one after. By the time your transaction confirms, you've received a slightly worse price. The searcher pocketed the difference. Nobody hacked you. The protocol worked exactly as designed.
That's MEV. Maximal extractable value, formerly miner extractable value. And when Ethereum moved from proof-of-work to proof-of-stake, the question of who captures that value shifted in ways that still aren't obvious to most people watching validator yields.
The short answer: specialist block builders now dominate proposal rights through a system called PBS (proposer-builder separation), so validators have largely stopped being the primary extractors. They've become landlords collecting rent. That sounds fine until you ask who sets the rent.
What MEV actually is, before the abstraction takes over
Forget the acronym for a moment. The core of it is ordering. Whoever decides the sequence of transactions inside a block controls whether a given opportunity is captured or left sitting.
The sandwich attack above is one flavour. Others: arbitrage between DEXs when prices drift apart, liquidation captures when a collateralised loan tips into insolvency, back-running a large trade to harvest the resulting price correction. Different shapes, same plumbing. They all depend on getting your transaction into the right position relative to someone else's.
In proof-of-work Ethereum, miners controlled ordering directly. In proof-of-stake, validators control it. Same logic, different actors. But here's where the market structure diverges hard.
How the builder market took over proposal rights
Validators are chosen pseudorandomly to propose each block, roughly one slot every 12 seconds. The naive picture is that the chosen validator assembles transactions from the mempool, sequences them to maximise fees, and submits the block.
Almost no validator actually does this.
The MEV landscape is too complex and too fast for most operators to navigate profitably on their own. Instead, an ecosystem of block builders competes to construct the most profitable block possible. A validator running MEV-Boost (the dominant relay middleware, built on Flashbots research) receives bids from multiple builders, picks the highest one, signs the block header, and collects a payment. The builder does the ordering work. The validator collects the cut.
The numbers are not trivial. Across measured periods, MEV-Boost has been active for well over 90% of Ethereum blocks, with the top three or four builders responsible for the majority of block production. This isn't a niche practice. It's the default. Think of it less as a feature and more as the actual water pressure behind the tap.
The rent model: what validators actually earn (and what they give up)
Run the scenario concretely. A block contains a rich arbitrage opportunity worth 0.8 ETH in extractable value, plus 0.05 ETH in priority tips. Base fees are burned under EIP-1559, so ignore those.
Without MEV-Boost, a solo validator with the skill and infrastructure to identify and sequence that arbitrage keeps the full 0.85 ETH. With MEV-Boost, a builder bids 0.6 ETH to the validator for the right to construct the block. The builder captures 0.2 ETH as margin, plus whatever additional value their superior searcher relationships extracted that the solo validator would have missed entirely.
So: 0.6 ETH instead of 0.85 ETH in this simplified comparison. But the validator also got there without running any MEV infrastructure. More to the point, if that solo validator realistically couldn't have captured the full 0.8 ETH on their own (and most can't), the builder's 0.6 ETH bid is pure upside over the 0.05 ETH baseline tip they'd have collected otherwise.
The core trade-off is this: validators exchange upside for reliability and simplicity. Builders absorb execution risk. Validators take predictable rent.
The deeper problem: when a few builders set the floor
This is where validator economics gets genuinely interesting, and where most casual analysis stops too early.
In a competitive builder market, builders bid aggressively to win proposal rights because losing means someone else captures the MEV. Competition should push bids toward the full value of the block, leaving builders with thin margins. For validators, that's ideal: near-full MEV capture without doing the work.
Builder markets, though, have strong centralising forces. Sophisticated order flow (private transaction feeds from wallets, DEX aggregators, and centralised exchanges) flows preferentially to builders who can offer better execution or rebates. A builder with exclusive order flow constructs more profitable blocks than one working purely from the public mempool. Over time, a small number of builders accumulate privileged relationships that become self-reinforcing. It's the same dynamic as a power grid where a few generators own the transmission lines.
When builder concentration is high, competitive pressure on bids weakens. Two builders don't bid against each other as ferociously as ten. The floor on validator payments drops. The margin captured by dominant builders widens.
Consider two validators with identical 32 ETH stakes. Maria runs MEV-Boost and connects to five relay networks, maximising her exposure to competing builder bids. She consistently captures 85-90% of available MEV value in each block she proposes. James runs MEV-Boost but connects to a single relay with limited builder competition. His bids run lower, and his effective annual yield ends up roughly 0.3-0.5 percentage points below Maria's, entirely because of relay and builder selection. Same hardware, same stake, different outcomes. That gap compounds over years into a meaningful ETH difference.
What people get wrong about MEV and validator centralisation
The common concern is that MEV creates pressure toward larger validators, since bigger operators can absorb infrastructure costs more easily and tilt the field against small stakers. That's real. It's also the second-order effect.
The first-order effect is subtler, and it cuts the other way: MEV-Boost actually democratised MEV access for small validators in the short term. Before PBS, a solo validator with 32 ETH had essentially no way to compete with mining pools running custom extraction software. MEV-Boost let that solo validator passively receive builder bids and collect meaningful MEV revenue without writing a line of specialised code. That was a genuine improvement.
The long-term risk is that as builder concentration increases, validators become price-takers in a market with limited sellers. The landlord analogy holds: if two landlords own all the desirable land, tenants still collect rent, but the landlords have pricing power.
There's also a trust assumption embedded in MEV-Boost that often goes unremarked, and it deserves more attention than it gets. The validator signs a block header before seeing the full block contents. They trust the builder (via the relay) to have included what was promised. If a builder equivocates or the relay behaves maliciously, the validator faces slashing risk. The relay is a trusted intermediary sitting inside a system built on trust-minimisation. That tension doesn't resolve itself cleanly, and anyone telling you otherwise is selling something.
Reading your validator's actual yield
If you're running a validator or evaluating liquid staking protocols, the figure to watch isn't the consensus layer APR alone (the base issuance reward for attesting and proposing). It's the execution layer APR, which captures MEV and priority fees. Beaconcha.in and similar explorers break this out clearly. If you're seeing execution layer rewards consistently close to or exceeding consensus layer rewards, your relay setup is working. If execution layer rewards are near zero, something in your relay configuration is wrong or absent.
Liquid staking protocols like Lido and Rocket Pool pass MEV revenue through to stakers, but the split between protocol, node operators, and stakers differs between them. That split matters more than the headline APR number on a landing page.
The builder market is not a footnote to Ethereum's validator economics. It's become the primary determinant of whether a validator earns 4% or 5% annually, whether small operators stay competitive, and whether block production remains meaningfully decentralised. Do you think those questions get resolved at the protocol layer, or in the commercial relationships between builders and order-flow providers? The base protocol is elegant. The MEV layer sitting on top of it is a market, with all the messiness markets produce: incumbent advantages, information asymmetries, and the slow accumulation of power toward whoever controls the most valuable order flow.
Validators didn't stop being important when builders arrived. The interesting work just moved somewhere else, and the rent checks kept coming.