Every time a transaction settles on a public blockchain, someone decides its position in the block. That sequencing decision has economic value. The aggregate of that value — known as maximal extractable value, or MEV — now runs into hundreds of millions of dollars annually across Ethereum alone, and the market structure that captures it is more sophisticated, and more concentrated, than most observers outside the space appreciate.

How the Market Is Structured

Post-merge Ethereum separated block proposal from block construction. Validators retain the right to propose; specialized builders compete to assemble the most profitable block and bid for that right through a relay layer, a model known as proposer-builder separation (PBS). MEV-Boost, the dominant middleware implementation, currently routes over 90 percent of Ethereum blocks through this system.

Below the builder layer sit searchers: automated agents that scan the mempool for extractable opportunities, primarily arbitrage between decentralized exchanges, liquidations, and sandwich positioning around large trades. Searchers submit bundles to builders; builders include the most profitable bundles and pass the net surplus to validators as a bid. The validator simply accepts the highest bid, often without visibility into what exactly generated it.

• Searchers compete on speed and algorithm quality, operating on latency measured in milliseconds.
• Builders compete on bundle inclusion and order-flow relationships, creating structural moats around proprietary deal flow.
• Validators receive a share of the surplus as an MEV boost on top of base issuance rewards.

Where Concentration Risk Accumulates

The builder market has consolidated materially. A small number of builders regularly construct the majority of blocks in any given week. This is partly a function of order flow: builders with exclusive or preferential access to large-volume platforms see richer bundles, which produce higher bids, which attract more validators, which reinforces their position. The dynamic is structurally similar to payment-for-order-flow arrangements familiar from traditional equities markets.

Private mempools and order flow auctions from decentralized applications have further stratified the ecosystem. Protocols that route user transactions directly to specific builders, bypassing the public mempool entirely, reduce sandwich exposure for users but effectively auction that flow to the highest bidder. The rents exist regardless of whether the mempool is public or private; the routing simply changes who captures them and when.

The Regulatory Surface

The structural parallels to traditional market microstructure are not lost on regulators. The SEC and CFTC have each signaled interest in how transaction ordering markets interact with existing market manipulation frameworks. The specific question under examination is whether certain MEV extraction strategies, particularly front-running and sandwiching, constitute manipulation under existing commodities or securities law if the underlying assets are deemed subject to those jurisdictions.

Simultaneously, the EU’s Markets in Crypto-Assets regulation introduces best-execution obligations for crypto-asset service providers. How those obligations interact with MEV dynamics in on-chain execution is an open interpretive question that compliance teams at regulated entities are actively working through. The regulatory surface here is not theoretical; it is live.

The Operator Read

For capital allocators observing this space, the MEV supply chain is worth mapping before making any infrastructure-adjacent allocation. The value capture question is structural, not speculative: block production economics favor entities with privileged order flow, and that structural advantage compounds over time. For operators building on-chain products that involve significant user transaction volume, the routing decisions made at the application layer have measurable economic consequences, both for users and for the protocol’s relationship with the builder market. Understanding the full stack, from mempool to validator, is baseline literacy at this point.