Maximal Extractable Value — MEV — was for several years one of the least understood and most consequential dynamics in blockchain ecosystems. The basic concept is that block producers (miners in proof-of-work systems, validators in proof-of-stake systems) have the ability to extract value from the transactions they include in blocks by ordering those transactions to capture arbitrage opportunities, front-run trades, sandwich user swaps, and exploit other patterns that depend on transaction ordering control. The value extracted through these activities has historically flowed to block producers and to the sophisticated trading firms that operate alongside them, often at the expense of ordinary users who pay implicit costs through worse execution prices on their transactions.
The MEV ecosystem in 2026 looks substantially different from the early days when MEV extraction was poorly understood and the tools to mitigate it were limited. Flashbots’ research and infrastructure development has produced visibility into MEV activity, mechanisms to allow users to protect their transactions from extraction, and an emerging architecture for redistributing MEV value to broader participants rather than concentrating it in a small set of professional extractors. The SUAVE project represents Flashbots’ most ambitious attempt to restructure MEV at the architectural level. The MEV-Boost relay system has become standard infrastructure for Ethereum validators. And the broader ecosystem of MEV-aware applications and infrastructure has matured significantly.
Understanding the state of MEV in 2026 requires looking at the actual mechanisms, the scale of value involved, and the various approaches that different participants are pursuing to address what is genuinely a complex multi-party problem.
The Scale of MEV and What It Represents
The total annual MEV extracted on Ethereum has been measured in the hundreds of millions to billions of dollars depending on the methodology and the market conditions. The vast majority of this extracted value occurs through three primary categories of activity: arbitrage (capturing price differences between trading venues, which is generally considered productive activity that improves market efficiency), sandwich attacks (placing transactions before and after a user’s trade to extract value from the price impact of the user’s trade, which is generally considered extractive), and liquidations (capturing value from undercollateralised positions on lending protocols, which is necessary infrastructure but has produced concentration concerns).
The composition of MEV activity matters because the appropriate response varies by category. Arbitrage MEV is genuinely productive and supports market efficiency; eliminating it would be counterproductive. Liquidation MEV is necessary for lending protocol operation but has concentrated to the point where a small number of sophisticated bots capture nearly all liquidation opportunities, raising questions about whether the value should be shared more broadly with the protocols that the liquidations support. Sandwich attack MEV is extractive at the expense of users and is the primary target of mitigation efforts.
The relative scale of these categories has shifted as the ecosystem has evolved. Sandwich attacks have declined as a share of total MEV activity because users have increasingly access to tools (private mempools, RPC services that protect from front-running, transaction batching) that reduce the surface area for sandwich extraction. Arbitrage MEV has remained substantial as the proliferation of trading venues and the introduction of new DeFi protocols has continued to produce arbitrage opportunities. Liquidation MEV scales with the total lending activity on DeFi protocols, which has grown significantly with the maturation of institutional DeFi credit markets.
MEV-Boost and the Builder-Relayer-Proposer Separation
The most consequential infrastructure development for MEV management on Ethereum has been the adoption of MEV-Boost and the proposer-builder separation architecture it implements. The mechanism splits the role of producing a block into three separate functions: searchers identify MEV opportunities and submit bundles of transactions that capture them; builders aggregate searcher bundles and other transactions into proposed blocks; relays connect builders to proposers and provide the trust layer that allows proposers to commit to blocks they have not directly constructed; proposers are the validators who actually publish blocks to the chain.
The architecture has several important properties. It separates the MEV extraction function (searchers and builders) from the consensus function (proposers), which prevents validator concentration from being driven primarily by MEV capability. It creates competitive markets at multiple levels — searchers compete for opportunities, builders compete for proposer attention, proposers select the most valuable blocks — which distributes MEV value across more participants than the pre-MEV-Boost architecture allowed. It provides transparency into MEV activity that supports research, mitigation development, and the broader understanding of the dynamics.
The adoption of MEV-Boost across Ethereum validators has been extensive, with most professional staking operations using the architecture. The result has been more visible MEV markets, more competitive distribution of MEV value, and a foundation for the next-generation MEV management architectures that the ecosystem is developing.
SUAVE and the Architectural Re-Imagination
SUAVE — Single Unifying Auction for Value Expression — is Flashbots’ most ambitious project, aiming to restructure MEV at the architectural level rather than addressing it through additional infrastructure layered on existing chains. The basic concept is that MEV management could be improved by separating the order flow auction function from any specific blockchain — creating a dedicated layer where transactions and intents from multiple chains can be aggregated, searchers can compete for opportunities across chains, and the value captured can be distributed back to users and protocols in ways that the per-chain architectures cannot support efficiently.
The strategic ambition is significant: SUAVE would not be a chain in the traditional sense but rather a coordination layer that handles MEV-related computation and value distribution while the underlying chains (Ethereum, L2s, other L1s) handle settlement. The architecture is technically complex and the deployment has been gradual, with various components launching and being tested before the full vision is realised.
The honest assessment of SUAVE’s progress is that the technical vision is compelling but the practical deployment requires substantial ecosystem coordination. Chains, applications, and users need to integrate with SUAVE for its value proposition to be realised, and the bootstrap problem for a coordination layer is genuinely difficult. The intellectual contribution of the SUAVE design has been significant — it has influenced how the broader MEV community thinks about the problem — even if the specific architecture’s commercial deployment is still developing.
The Application-Level Mitigations
Beyond the infrastructure-level MEV management, application-level mitigations have become increasingly common. Decentralised exchanges have implemented various mechanisms to reduce sandwich attack vulnerability: Hyperliquid’s order book architecture avoids the AMM dynamics that produce sandwich opportunities; CoW Protocol (CoWSwap) batches user swap intents to find matching opportunities that avoid the impact of individual trades; UniswapX uses a Dutch auction mechanism that lets searchers compete to provide best execution to users.
The aggregate effect of these application-level mitigations has been to reduce the user-extractive MEV by giving users access to better execution tools. The mitigation works best for users who explicitly use the protected venues; users transacting directly on AMMs without intent-based protection remain exposed to sandwich extraction.
The Layer 2 ecosystem has been more proactive about MEV management than Ethereum mainnet because L2 sequencer operators have the ability to implement custom transaction ordering rules. Several L2s have implemented first-come-first-serve sequencing, encrypted mempools, or other mechanisms that reduce the surface area for extractive MEV. The variation across L2s has produced an environment where users seeking MEV protection can choose L2s with stronger protections, which creates competitive pressure on other L2s to improve their MEV management.
The Distribution and Redistribution Question
The most consequential ongoing debate in the MEV community is about how the value captured through MEV should be distributed. The pre-MEV-Boost default was that block producers captured most of the value, with sophisticated searchers extracting some share. The MEV-Boost architecture distributed value more broadly across the searcher-builder-proposer stack. The application-level mitigations have shifted value back to users by reducing extractive opportunities.
The next phase of redistribution involves protocols capturing value that is currently captured by external searchers. Liquidation auctions on Aave and Morpho could in principle direct the liquidation premium to protocol revenue rather than to external liquidators. DEX-based arbitrage value could in principle flow to DEX protocols rather than to external arbitrageurs. The technical mechanisms for these redistributions are developing but require careful design to avoid creating perverse incentives or breaking the productive functions that some MEV activity supports.
The validator dimension of redistribution is also important. Institutional Ethereum staking increasingly views MEV revenue as a meaningful component of staking yield, and the distribution of MEV value across validators affects the economics of professional staking operations. The proposer-builder separation has helped equalise this distribution, but ongoing changes to MEV management infrastructure continue to affect how the value flows.
The Honest Assessment for Different Participants
For end users, the MEV environment in 2026 is meaningfully better than the environment of three years ago. Users who transact through MEV-aware venues (CoWSwap, UniswapX, protected L2s) face significantly less extractive MEV than users transacting on raw AMMs or unprotected mainnet venues. The available tools provide a real protective benefit that did not exist in earlier MEV environments.
For protocol developers, MEV considerations have become standard design inputs. New DeFi protocols are designed with MEV awareness as a baseline requirement, and the protocols that fail to account for MEV in their architecture face competitive disadvantage from those that do. The intellectual maturity of MEV-aware protocol design has compounded into substantially better outcomes for the ecosystem.
For institutional participants, MEV transparency and management has been one of the requirements for sustained institutional engagement with DeFi. Institutions cannot accept the opacity and extractive risk that early DeFi posed; the infrastructure improvements that Flashbots and the broader MEV community have produced have been preconditions for institutional comfort with on-chain activity.
The honest position is that MEV is and will remain a permanent feature of blockchain ecosystems — the underlying dynamic of value capture from transaction ordering control is intrinsic to how blockchain consensus works. The question is not whether MEV exists but how it is captured, distributed, and made transparent. The progress of the past several years has been substantial, the projects working on the problem are technically sophisticated, and the trajectory continues to improve outcomes for users and protocols at the expense of pure extraction. The work is not finished, but the direction is clearly favourable, and the contrast with the early MEV environment is one of the most concrete examples of how blockchain infrastructure can mature when serious technical and economic work is sustained over time.
Who Actually Benefits From MEV — and Who Is Pretending Not to Notice
Let us be precise about something the redistribution framing consistently obscures. MEV is not an abstract inefficiency that smart protocol design can gradually engineer away. It is a transfer — value moving from one set of participants to another through mechanisms that the losing parties never consented to and often cannot detect. The builders who captured sandwich attack revenue over the past three years did not earn that value through productive work. They extracted it from users who thought they were getting market prices and were not. The fact that Flashbots has made this extraction more visible and introduced competition into the extraction pipeline is a meaningful improvement. It is not the same as accountability.
The question any honest analyst should ask is: who had the power to extract MEV before MEV-Boost existed, who has it now, and how has the structure of that power shifted? The answer is that the power was concentrated in block producers and in a small number of sophisticated trading operations with co-location access to mempool data. MEV-Boost spread some of that concentration into a builder market with more participants competing. The direction is right. But the builders who dominate block construction today — a small number of highly capitalised operations — still capture the majority of MEV value through their position in the supply chain. The language of “redistribution” implies that a meaningful share of this value is flowing back to ordinary users. The data suggests the share is real but modest.
The regulatory dimension of MEV is also more consequential than it appears in most ecosystem commentary. Regulators in the European Union — working through the frameworks established by MiCA and the provisions of the EU AI Act’s high-risk AI compliance framework — have begun to ask whether automated trading systems that systematically extract value from other market participants through information asymmetry constitute a form of market manipulation under existing securities and market integrity law. The answer is not settled. The question is legitimate, and the MEV community has largely chosen not to engage with it directly.
The uncomfortable accountability framing for the MEV ecosystem is this: if a system were designed from scratch to allow sophisticated participants to systematically profit from information advantages that ordinary users cannot access, and if the description of that system were placed in a regulatory filing rather than a technical whitepaper, it would face serious questions about investor protection and market integrity. The technical sophistication of the MEV infrastructure does not change the underlying power structure — it merely makes the extraction more efficient, more opaque to non-technical observers, and harder to regulate because the mechanisms evolve faster than the regulatory frameworks can track them. Progress on MEV is real. Honest accounting of who still benefits and who still pays is rarer than the progress deserves.
The Invisible Extraction: A Field Report on Who Profits From MEV and What SUAVE Actually Changes
Begin with the concrete particular. A trader submits a transaction on Uniswap to buy 10 ETH. The transaction sits in the public mempool for the 12 seconds between Ethereum blocks. In that window, a searcher’s automated system detects the pending transaction, calculates that the trade is large enough to move the market price by 0.3%, and submits two of its own transactions: one that buys ETH just before the user’s trade (pushing the price up), and one that sells immediately after (capturing the price the user paid above market). The user’s 10 ETH costs more than it should have. The searcher captures the difference. No law was broken. No contract was violated. The mechanism that allowed it is a design choice: in Ethereum’s architecture, the order of transactions within a block is determined by the block producer, not by time of submission.
This is maximal extractable value. The “maximal” in the name replaced the original “miner” after the Merge — validators now do what miners once did — but the extraction mechanism is unchanged. And the scale is not trivial. Flashbots’ public MEV data shows cumulative extraction across Ethereum of well over $1 billion since the concept was first quantified. The per-day extraction rate in 2026 has moderated from the 2021–2022 peaks but remains in the range of $3–8 million per day in active DeFi markets. Per year, that is $1–3 billion extracted from users of decentralised exchanges, arbitrage-sensitive protocols, and liquidation mechanisms — transferred to searchers, block builders, and validators.
Flashbots was founded in 2020 with the explicit goal of making MEV extraction more transparent and less harmful. Its SUAVE architecture — in production testing through 2025–2026 — is a decentralised block-building protocol designed to solve the worst MEV pathologies: competitive gas auctions that clogged the network, sandwich attacks that cost retail users money, and information asymmetry that concentrated extraction in a small number of sophisticated actors. In principle, SUAVE creates a marketplace where users can express preferences about transaction ordering and searchers can bid on bundles in ways that share value with users rather than extracting from them.
The mechanism is genuinely more sophisticated than pre-Flashbots chaos. Ethereum’s Pectra account abstraction changes have created additional tooling for users to specify transaction conditions that reduce vulnerability to certain MEV patterns. The competitive marketplace has distributed some of the MEV rent to validators that previously went to mining pools. L2 sequencer revenue and who captures it shows what centralised sequencer control looks like as a comparison: L2 sequencers capture MEV by virtue of controlling transaction ordering without any of the competitive auction mechanics that Ethereum mainnet has developed.
What Flashbots and SUAVE have not changed is the information asymmetry. The retail user who submits a Uniswap transaction does not have the mempool monitoring infrastructure, the co-located servers, or the capital to participate in the MEV auction as a searcher. The SUAVE competitive marketplace is competitive among the participants who can actually compete — a subset of sophisticated actors with the same structural advantages searchers had before Flashbots. The redistribution from searchers to validators is real. The redistribution from extraction to user protection is modest.
Solana’s DeFi ecosystem post-ETF operates under different MEV mechanics — Solana’s local fee markets and shorter block times reduce certain MEV vectors — but the underlying principle holds: wherever there is a public ordering mechanism and a price-sensitive transaction, there is extractable value. Solana’s institutional adoption trajectory has brought new capital into the DeFi ecosystem without resolving the MEV question for retail participants on either chain.
how regulatory rulemaking captures market structure illustrates the same dynamic: regulatory frameworks can formalise market structure without necessarily changing who benefits from it. MEV is not a bug in the Ethereum protocol. It is a consequence of building a permissionless public system where transaction order determines profit. SUAVE has made the extraction more transparent, more competitive, and somewhat more shared. The retail user who submits a Uniswap transaction in 2026 is paying less to MEV than in 2021. That is progress. It is not protection. The structural information advantage that sophisticated searchers hold is not resolved by better auction mechanics; it is expressed through them.
