MEV on Ethereum: Quantifying the Invisible Tax

MEV on Ethereum: Quantifying the Invisible Tax

A quantitative study of Maximal Extractable Value on Ethereum, measuring the economic impact on end users and evaluating the effectiveness of mitigation mechanisms including Flashbots, MEV-Share, and the evolving proposer-builder separation architecture.

James ParkFebruary 15, 2026
15 min

Maximal Extractable Value (MEV) — the profit that can be extracted by reordering, inserting, or censoring transactions within a block — has been one of the most consequential and least understood phenomena in Ethereum's ecosystem. This study quantifies MEV extraction on Ethereum from January 2025 through February 2026, analyzing over 18 million blocks and 4.2 billion transactions to measure the total economic value extracted, the distribution of that value among ecosystem participants, and the effectiveness of mitigation mechanisms. Our findings indicate that total MEV extraction during this period exceeded $2.1 billion, with approximately $780 million representing a direct cost to end users through worse execution prices.

The composition of MEV has shifted significantly since the early days of Flashbots. Sandwich attacks — where a searcher places orders before and after a victim's trade to profit from the price impact — accounted for 42% of total extracted value in our study period, followed by arbitrage (35%), liquidations (18%), and other strategies (5%). Notably, the absolute volume of sandwich attacks has decreased from its 2023 peak, partly due to improved DEX router designs that minimize slippage and partly due to the migration of trading volume to private order flow channels. However, the sophistication of remaining attacks has increased, with cross-domain MEV strategies that exploit price discrepancies across L1 and multiple L2s now accounting for 15% of all sandwich profits.

The Flashbots-developed MEV-Share protocol, launched in 2023, was designed to return a portion of MEV profits to the users whose transactions generated them. Our analysis of MEV-Share transactions shows that the protocol has returned approximately $180 million to users over our study period, representing roughly 23% of the MEV generated from MEV-Share order flow. While this is a meaningful improvement over the zero rebate that users received historically, the majority of MEV value still flows to searchers and block builders. The adoption of MEV-Share has been uneven: MetaMask integrated the protocol for all transactions by default, but many other wallets and applications have not, creating a two-tier system where sophisticated users capture rebates while others continue to bear the full cost of MEV extraction.

Proposer-Builder Separation (PBS), implemented through MEV-Boost, has become nearly universal on Ethereum, with over 95% of blocks built through the builder-proposer marketplace. The builder market has consolidated significantly: the top three builders (Beaver Build, Flashbots Builder, and Titan Builder) produce approximately 80% of all blocks, raising concerns about centralization and potential censorship. Our data shows that during the OFAC sanctions compliance period, builders that filtered transactions related to sanctioned addresses produced 60-70% of blocks, resulting in measurable delays for affected transactions but not outright censorship, as non-filtering builders eventually included them. The upcoming ePBS (enshrined PBS) upgrade to Ethereum aims to reduce reliance on external relay infrastructure, though its timeline remains uncertain.

The economic impact of MEV on Layer 2 networks requires separate analysis, as the MEV dynamics differ substantially from L1. On L2s with centralized sequencers (which includes all major rollups as of this writing), the sequencer operator has exclusive authority over transaction ordering, creating a monopolistic MEV extraction opportunity. Our data indicates that L2 sequencers currently extract minimal MEV from users — primarily because the operators are reputable entities (Offchain Labs for Arbitrum, OP Labs for Optimism, Coinbase for Base) with reputational incentives to avoid predatory behavior. However, as L2s decentralize their sequencers, the same MEV dynamics observed on L1 will likely emerge, and the mitigation infrastructure will need to be adapted for the multi-chain environment.

Several promising approaches to MEV mitigation are in development. Encrypted mempools, where transactions are encrypted until block ordering is finalized, could eliminate frontrunning and sandwich attacks entirely. Threshold encryption schemes from Shutter Network and Chainlink's Fair Sequencing Services are in testnet, though they introduce latency overhead that may be unacceptable for time-sensitive applications. Order flow auctions, where the right to execute a user's trade is auctioned competitively, could ensure users receive the best available execution price. UniswapX's auction-based routing has demonstrated this approach at scale, processing over $40 billion in volume with measurably better execution than traditional AMM swaps. The long-term equilibrium likely involves a combination of these approaches, with different mechanisms suited to different transaction types and urgency levels.

Discussion

Loading comments...