Ethereum's block production pipeline has consolidated into a three-firm oligopoly. As of April 2, 2026, relayscan.io data shows Titan Builder commanding 52.16% of MEV-Boost delivered blocks over a 24-hour window, followed by Flashbots' BuilderNet at 24.63% and Quasar at 15.06%. Those three entiti...
Ethereum's block production pipeline has consolidated into a three-firm oligopoly. As of April 2, 2026, relayscan.io data shows Titan Builder commanding 52.16% of MEV-Boost delivered blocks over a 24-hour window, followed by Flashbots' BuilderNet at 24.63% and Quasar at 15.06%. Those three entities account for roughly 96% of block construction on a network that processes over $200 billion in stablecoin settlement per month and secures $340 billion in total value across Layer-1 and Layer-2 protocols.
The relay layer — the intermediary that passes builder-produced blocks to validators — has undergone a parallel consolidation. Ultrasound Relay leads at 33.92% of delivered payloads, with Titan Relay at 24.19%, bloXroute Max-Profit at 14.67% and Aestus at 10.03%. Flashbots' historical flagship relay, boost-relay.flashbots.net, has collapsed to 4.22% of market share, a reversal from the 70%-plus dominance it held in 2022.
The concentration is a direct consequence of exclusive order flow (EOF) economics. Builders with private deal flow — Titan via its Banana Gun agreement, BuilderNet via the Flashbots searcher ecosystem — can construct higher-value blocks than open-market competitors. Validators, acting rationally, route to whichever relay serves the highest-paying block. The result is a winner-take-most dynamic that the Ethereum Foundation now treats as a protocol-level concern. On March 2, 2026, Vitalik Buterin published a multi-stage plan centered on enshrined Proposer-Builder Separation (ePBS), Fork-Choice Enforced Inclusion Lists (FOCIL) and encrypted mempools, targeted for the Glamsterdam and Hegota upgrades in 2026.
This report quantifies the current market structure, examines the revenue that flows through each layer of the MEV supply chain, and evaluates whether the proposed protocol changes can credibly restore competition.
Ethereum block construction is handled almost entirely off-protocol through MEV-Boost, the Flashbots software that implemented Proposer-Builder Separation in September 2022. Builders assemble blocks by combining public mempool transactions, private order flow and MEV bundles, then submit bids to relays, which forward the highest bid to validators.
According to relayscan.io data snapshot for April 2, 2026, the 24-hour builder league table stands as follows:
On a 7-day average, Titan holds 46.77% and BuilderNet 39.19%, with Quasar and Beaverbuild splitting most of the remainder. The concentration exceeds the 75% three-builder share cited in protocol research papers as of late 2025, and is higher than the 86% two-firm duopoly (Beaverbuild and Titan) measured in March 2025 before BuilderNet's production ramp.
Titan's rise from under 1% market share in early 2023 to majority position is the defining shift of the cycle. The firm entered into an exclusive order flow agreement with Banana Gun, the Telegram trading bot that became a dominant retail-order aggregator on Ethereum and Base. That deal gave Titan access to a stream of high-MEV transactions — sandwich-prone swaps, newly launched token trades, arbitrage opportunities — that no other builder could replicate. Titan's on-chain revenue from block building surged past $34.5 million in a single week following a $50 million trader error that Titan's blocks captured.
BuilderNet, launched by Flashbots on November 26, 2024, represents a structural counter-design. Rather than operate as a single centralized builder, BuilderNet runs a multi-operator system inside trusted execution environments (TEEs). Initial partners are Flashbots, Beaverbuild and Nethermind. On December 8, 2024, Flashbots deprecated its standalone centralized builder and migrated all order flow to BuilderNet. Version 1.2, shipped February 2025, streamlined operator onboarding. As of April 2026, BuilderNet has captured the number-two slot but has not unseated Titan.
Quasar, a newer entrant, has climbed to 15% market share through a combination of private searcher integrations and aggressive rebate strategies. Its profitability per block remains below Titan's, but its share is growing.
Relays sit between builders and validators. They verify builder bids, hold blocks until validators commit to them, and forward the winning payload. There are no relay fees; relays operate at a loss and are subsidized by the operating entities as a public good or loss-leader for other business lines.
The April 2, 2026 relay league table (24-hour payload share) from relayscan.io:
Flashbots' historical relay, which processed a majority of MEV-Boost blocks from late 2022 through mid-2023, now handles barely one in every 24 blocks. The decline tracks two structural shifts. First, Flashbots' relay enforces OFAC compliance by default, filtering transactions involving sanctioned addresses such as those derived from Tornado Cash. Non-censoring relays — Ultrasound, Agnostic, Aestus — captured validator demand from operators unwilling to accept compliance filtering. Second, Titan launched its own relay (titanrelay.xyz) to capture the rent that its builder dominance generates, cutting Flashbots out of the Titan order flow entirely.
Over 70% of Ethereum blocks have been OFAC-compliant at some point post-Merge, driven primarily by builder and relay filtering rather than validator preferences. The current four-relay dominance — Ultrasound, Titan, bloXroute, Aestus — collectively handles over 80% of all payloads.
The centralization is not an accident of early adoption. It is the equilibrium outcome of a market where differentiation is produced by private order flow rather than by technical skill.
Academic research published in the Advances in Financial Technologies 2024 proceedings describes a "chicken-and-egg" dynamic: builders need differentiated order flow to win auctions, but searchers and order-flow providers only route to builders with significant market share because that maximizes inclusion probability. Once a builder crosses a share threshold — empirically around 20% — exclusive deals become economically viable and reinforce the position.
Titan's Banana Gun agreement is the canonical example. Beaverbuild-rsync's integrations with Wintermute and SCP searchers are a second. BuilderNet attempts to short-circuit the logic by pooling order flow across multiple operators inside a TEE, so that no single operator has private visibility, and by open-sourcing refund rules. But even BuilderNet depends on Flashbots' historical searcher relationships for its order flow pipeline.
The outcome is that roughly 96% of Ethereum blocks are produced by three legal entities, each of which can theoretically refuse service to specific addresses, front-run trades, or coordinate with regulators. The Ethereum Foundation has published research flagging this as the highest-severity centralization risk on the network.
The full MEV economy on Ethereum is estimated at $1 billion to $2 billion annually in extracted value, with upper-range estimates from Flashbots' own research placing total MEV (including non-extracted opportunity value) at $3 billion to $7 billion per year. That figure compares to approximately $65 million in annual Ethereum base-layer fees and roughly $4 to $5 billion in staking issuance.
Within the MEV pool, value distribution runs roughly as follows:
For Titan specifically, the firm's block-building gross revenue during peak weeks has exceeded $34 million, implying an annualized run-rate in the high hundreds of millions if peaks were sustained. Actual annualized margin is almost certainly lower — $50 million to $150 million is a reasonable range — but it is sufficient to sustain engineering headcount, exclusive-deal payments to order-flow providers and competitive co-location infrastructure.
The economics favor monopoly. A builder with 50% market share has twice the auction participation of a 25% competitor, which means higher hit rates on exclusive order flow, which means more MEV captured per block, which means higher validator payments, which means more validators route to that builder's relay. The feedback loop is the reason the market has consolidated rather than fragmented despite low technical barriers to entry.
On March 2, 2026, Vitalik Buterin published a multi-stage plan to reduce block builder centralization, scheduled across the Glamsterdam and Hegota upgrades.
Enshrined Proposer-Builder Separation (ePBS): Glamsterdam, expected in the first half of 2026, will move the PBS mechanism from the off-protocol Flashbots software into the Ethereum consensus layer itself. Validators will natively outsource block building to a permissionless builder market, eliminating the need for MEV-Boost and its relay intermediaries. Buterin has publicly cautioned that ePBS alone does not eliminate centralization, because a competitive builder market can still consolidate around exclusive order flow.
Fork-Choice Enforced Inclusion Lists (FOCIL): Scheduled as the headline feature of the Hegota upgrade in the second half of 2026, FOCIL allows a small randomly-selected set of validators to enforce the inclusion of specific transactions in the next block. A block that omits a FOCIL-mandated transaction is rejected by the fork choice rule. The design provides a censorship backstop even if a single hostile builder controls 100% of block production.
Big FOCIL: Buterin has also floated expanding the mechanism so that FOCIL participants collectively assemble the entire transaction list for each block, leaving builders responsible only for ordering MEV-sensitive transactions and computing state. This would effectively commoditize block building and strip pricing power from incumbents.
Encrypted Mempools: To address toxic MEV — sandwich attacks and front-running that rely on visibility into pending transactions — Buterin endorses encrypting transactions until block inclusion, using threshold cryptography or TEEs. Several research teams have working prototypes, but no production deployment has been scheduled.
None of these mechanisms are live in April 2026. Glamsterdam timing has slipped from original 2025 projections, and Hegota is dependent on Glamsterdam shipping first. Execution risk is material.
The three-firm builder oligopoly creates a single point of regulatory leverage. If the U.S. Treasury issues OFAC guidance targeting a specific address, Titan, BuilderNet and Quasar can comply within hours — as Flashbots' relay did after the Tornado Cash sanctions in August 2022. A Ukrainian sanctions list, EU MiCA enforcement action, or U.S. court order could similarly propagate through the network at the speed of a builder software push.
The mevwatch.info data, which tracks OFAC-compliance of recent blocks, has shown censorship rates ranging from 30% to 80% depending on which builders dominate at a given time. The April 2026 ratio is reportedly near the lower end of that range, reflecting Ultrasound Relay's non-censoring stance, but the structural vulnerability remains.
Wahrstätter's October 2023 observation — that two builders accounted for 88.7% of blocks — has worsened in absolute concentration and improved only because one of those two builders (BuilderNet, successor to Flashbots' builder) adopted TEE-based multi-operator architecture. The improvement is architectural, not market-share-based.
Ethereum's block production layer is now more concentrated than at any point since Proposer-Builder Separation was introduced. Three firms construct 96% of blocks. One firm, Titan, holds majority share. The relay tier has shifted away from the Flashbots-centric architecture of 2022-2023 toward a four-firm distribution led by Ultrasound and Titan's own captive relay. Flashbots' own relay is a rounding error.
The cause is not technical failure. MEV-Boost works as designed. The cause is that exclusive order flow generates compounding returns to scale, and no countervailing market force exists below the protocol layer. BuilderNet is the most credible market-based counter-design, and it has reached 24.63% share, but has not reversed the trend.
The Ethereum Foundation's response — ePBS, FOCIL, Big FOCIL, encrypted mempools — places the centralization problem on the protocol roadmap rather than leaving it to market dynamics. The 2026 upgrade calendar is the critical test. If Glamsterdam ships on schedule with ePBS, and Hegota delivers FOCIL, the centralization pressure moves from the application layer to the consensus layer, where it is constrained by the validator set rather than by private commercial agreements. If the upgrades slip, the three-firm oligopoly persists and the censorship-by-default risk persists with it.
Economically, the MEV supply chain is a useful lens on the broader webthreepedia thesis that blockchain value flows are dominated by subsidy mechanisms and concentrated rent extraction rather than by transparent on-chain fee markets. The $1-7 billion MEV pool dwarfs the $65 million Ethereum base-layer fee take by two orders of magnitude. The validators who benefit are diffuse; the builders who broker the value are not. Until the protocol absorbs the function that Titan currently monopolizes, that imbalance is structural.