MEV-Boost Relays: The Middlemen of Block Construction
The MEV-Boost relay protocol is the backbone of Ethereum's current proposer-builder separation (PBS) implementation. These relays act as the middlemen between block builders and validators, receiving sealed block bids, selecting the highest-paying payload, and forwarding it to the proposer for inclusion. Without relays, builders would have to broadcast full blocks to every validator, leaking MEV-rich transactions and enabling frontrunning. Relays bundle the economic incentives of MEV extraction with the security requirements of block production, creating a marketplace where builders compete for inclusion in the next slot.
This guide dissects the inner mechanics of MEV-Boost relays, their trust models, and their profound effect on MEV extraction. We’ll explore how relays act as gatekeepers of builder competition, the trade-offs between latency and censorship resistance, and the emerging risks of relay centralization. Whether you’re a validator running mev-boost or a builder optimizing submission strategies, understanding the relay layer is essential to navigating Ethereum’s evolving block construction landscape.
- MEV-Boost relays enable a blind auction where builders submit sealed blocks; the relay selects the highest bid and forwards the header to the validator.
- Relays introduce trust assumptions: they can censor, frontrun, or withhold data, though economic incentives keep most honest.
- Multi-relay setups reduce trust concentration but increase latency; a trade-off every validator must consider.
- Builder competition inside relays drives down validator bids but also creates MEV leakage to the relay layer.
- Real relay incidents (e.g., Flashbots outage, Eden shutdown) highlight the importance of relay diversity and monitoring.
- The future lies in ePBS (enshrined PBS) and relay-less designs that eliminate the middleman, but relays will remain critical for several years.
What is an MEV-Boost Relay? The Gatekeeper of Blinded Blocks
An MEV-Boost relay is a server that intermediates the block auction between builders and validators. When a builder constructs a block with MEV-extracted transactions, it cannot reveal the full block before auction—doing so would let validators or competitors steal the profitable order flow. Instead, builders send a blinded block to the relay: a block header containing only the commitment (hash) of the full payload and a bid (e.g., in ETH or native token). The relay aggregates these blinded bids from multiple builders, selects the highest bid, and forwards that header to the validator. The validator signs the header and returns it to the relay; only then does the relay reveal the full block body to the validator for inclusion.
This design was introduced by Flashbots in 2022 as an out-of-protocol solution to combat time-bandit attacks and uniform block value leakage. Today, multiple relay providers operate: Flashbots (centralized relay), Bloxroute (max-profit and ethical relays), Titan (with native block builder), Ultra Sound, Agnostic, and others. Each exposes an /eth/v1/builder/ API following the MEV-Boost spec, but their trust models and fee structures differ. The relay’s role is purely administrative: it does not select transactions or build blocks, but it controls which builder wins the auction and when the final block is signed.
How PBS Created the Relay Layer: From Mempool Auctions to Slot Auctions
Before PBS, validators built blocks themselves by scanning the public mempool for profitable transactions. This led to centralization—only sophisticated validators with custom MEV strategies could capture full value. The introduction of proposer-builder separation via MEV-Boost externalized block building to specialized builders. However, a direct builder-to-validator connection introduces security risks: if builders broadcast blocks prematurely, front-runners can steal MEV. Relays solve this by acting as a blind auctioneer.
The relay layer emerged from the MEV-Boost middleware written by Flashbots. It connects to the validator engine API and exposes a builder API. The relay stores a mapping of validator registrations (public keys) and relay configurations. Each slot, the validator requests a header from the relay (or multiple relays in multi-relay setups), signs the best header, and sends the signature back. The relay then delivers the full block body. This design isolates the validator from the builder’s contents until after signing, preventing transaction censorship by the validator itself (though not by the relay).
Ethereum’s transition to proof-of-stake magnified the need for relays because validators now propose blocks only rarely (any given validator is selected infrequently); missing a high-value block due to poor bidding strategy is costly. Relays enable a global marketplace where builders from any location can compete for every slot, increasing aggregate MEV extraction efficiency.
Trust Models: Single-Relay vs. Multi-Relay and the Risk of Censorship
Every MEV-Boost relay introduces a trust assumption: the relay must be honest in two respects. First, it must correctly forward the winning bid’s header to the validator (no injection of its own block). Second, it must reveal the full block body after receiving the validator’s signature (no data withholding). If a relay is malicious, it can censor builders (by ignoring their bids), censor transactions (by forwarding only blocks that exclude certain addresses), or steal MEV (by simulating the winning block and frontrunning it itself).
Validators can run a single-relay configuration, pointing mev-boost to one relay endpoint, or a multi-relay configuration, using a list of relays. In multi-relay mode, the validator fetches headers from all relays simultaneously and selects the highest bid among them. This reduces trust in any single relay because the validator sees multiple options. However, it increases latency—each relay must be queried, and the validator must wait for all responses before choosing. Real-world setups often use a custom relay selector (e.g., mev-boost with --relays flag) that fetches in parallel and cancels slow relays.
The trade-off is stark: faster relays offer lower latency but higher censorship risk. For example, Flashbots’ relay historically had a compliant filter that censored transactions interacting with certain sanctioned addresses, while Bloxroute’s “max-profit” relay allows any transaction but may charge higher fees. Advanced validators can weight relays by reputation and fee schedule, but no mechanical solution removes the need for trust entirely—hence the push toward ePBS (enshrined PBS) that will eliminate relays altogether.
Builder Competition Inside the Relay: How Bids Are Evaluated and What Determines the Winner
When a builder submits a blinded block to a relay, it includes a bid—a payment to the validator in ETH or native token. The relay evaluates bids from all builders for the current slot and selects the highest bid. Most major relays (such as Flashbots, Ultra Sound, and Agnostic) run at no fee and forward the full bid to the validator, though some relay operators have experimented with fee or revenue-sharing arrangements. Builders must still ensure their block is valid; otherwise, the relay will reject the submission.
Builder strategy revolves around estimating the bids of competitors. Sophisticated builders use game theory: they bid just enough to win, knowing that other builders are also optimizing. The relay acts as an oracle for validator demand; if validators are scarce (e.g., low number of active validators per slot), bids can be lower. Relays also validate submissions to filter out spam and invalid blocks.
Builder competition drives down validator profits in the long run, as any excess rent is competed away. Relays therefore create a two-sided market: builders compete for access to the validator via the relay. Because most major relays run at no fee, this competition primarily benefits validators through higher bids rather than the relay itself, though the relay’s gatekeeping role still gives it significant influence over which builders win.
Relay Design Trade-offs: Latency vs. Censorship Resistance vs. Profit
Every relay must balance three design dimensions: latency, censorship resistance, and profit incentives. Low-latency relays (e.g., those geographically close to validator clients) can reduce the time between header fetch and block signing, which is critical for validators in competitive slots. However, low-latency often requires centralized infrastructure, making it easier for a relay to censor builders or transactions. High-latency relays (e.g., those with multiple redundant nodes) are more censorship-resistant but may cause the validator to miss the slot deadline, resulting in a missed proposal and penalty.
Fee and incentive models also vary. Most major relays (such as Flashbots, Ultra Sound, and Agnostic) run for free as public goods, while some relays have been operated by builders themselves or with revenue-sharing arrangements. These differences influence which builders participate and how aggressively they bid. Some relays, such as Bloxroute’s max-profit relay, are designed to maximize validator payout, while others prioritize simplicity or censorship policy over raw profit.
The table below summarizes key differences among popular relays:
How Relays Affect MEV Extraction: Information Gatekeepers and Block Content Selection
Relays do not directly extract MEV—they don’t reorder or include transactions—but they profoundly shape which MEV opportunities are captured. By choosing which builder’s bid wins, the relay determines whose block gets included. If a relay favors a particular builder (e.g., because of low latency or a prior relationship), that builder can consistently win slots and capture most MEV in the ecosystem. This can lead to builder centralization: top builders like Beaver, Titan, and Rsync often dominate because they have preferred relay partnerships.
Moreover, relays see the full block before the validator does. A malicious relay could simulate the block to identify profitable arbitrage opportunities and frontrun the builder in a subsequent slot. This is a form of “relay MEV theft,” which the MEV-Boost protocol mitigates by requiring the relay to commit to the block hash before revealing the body. But if the relay can revert its commitment (by refusing to release the block after signature), it could still extract value. In practice, reputable relays are economically incentivized to be honest because cheating would destroy their reputation and reduce validator adoption.
The relay’s position as a middleman also affects the distribution of MEV between validators and builders. Validators receive the bid, while builders keep any surplus above what they bid. Because most major relays operate at no fee, the relay layer itself typically captures little to no value directly; the primary concern is not relay fees but the trust, censorship, and centralization risks relays introduce.
Case Studies: Real-World Relay Incidents and What They Teach Us
In 2023, the Flashbots relay experienced bugs and downtime that caused affected validators to miss MEV-boosted blocks for a number of slots. Validators using only Flashbots as their relay missed revenue, while those with multi-relay fallbacks could switch to Bloxroute or others. The incident demonstrated the fragility of single-relay dependency and spurred adoption of relay diversification and monitoring practices.
Another crucial event was the Eden Network’s relay shutdown in 2022. Eden operated a relay that prioritized its own builder, leading to a conflict of interest. The community reacted negatively, and validator adoption plummeted. Eden eventually pivoted away from relay operation. This highlighted the trust requirement that relays must be neutral; any hint of builder favoritism erodes validator trust.
More recently, Bloxroute’s ethical relay implemented an OFAC filter, sparking debate about mandatory censorship in relays. Some validators switched to the permissive max-profit relay to avoid censorship. These case studies reveal that relays are not merely technical components but also political and economic actors. Validators must continuously evaluate relay trustworthiness and adjust configurations accordingly.
Risks and Mitigations: Relay Centralization, Data Withholding, and Future Directions
The most pressing risk in the current relay ecosystem is centralization. A large share of MEV-Boost blocks is produced through just a handful of relays (such as Flashbots, Ultra Sound, Bloxroute, and Titan). This concentration creates a single point of failure for censorship or network attacks. If a relay coalition decided to censor certain transactions, they could effectively control what is included in a majority of blocks. Furthermore, relay operators are typically US-based entities subject to OFAC regulations, increasing the compliance risk.
To mitigate these risks, several approaches are being developed. Multi-relay clients allow validators to query many relays simultaneously, reducing dependence. Relay reputation systems (e.g., relay monitoring dashboards) let validators choose based on past performance. The Ethereum core community is actively working on enshrined PBS (ePBS), which would codify the relay-like functionality directly into the consensus protocol, eliminating the need for trusted third parties. ePBS is still in research but is scheduled for a future hard fork.
Another mitigation is relay-agnostic builder submission via direct validator APIs bypassing relays altogether. Such approaches would let builders broadcast blinded blocks to validators over a peer-to-peer network, removing the relay as a necessary intermediary. However, such designs require significant protocol changes and are not yet deployed.
The Future of MEV-Boost Relays: From Middlemen to Optional Infrastructure
The long-term trajectory suggests that relays will become less central as Ethereum moves toward ePBS and other relay-less designs. In an ePBS world, validators would run a built-in block proposer selection mechanism that automatically picks the highest bid from builders without a relay. This removes the trust assumption entirely and reduces MEV leakage. However, ePBS is years away; until then, relays remain the dominant infrastructure for MEV extraction.
In the near term, we may see relay consortiums that spread trust across multiple operators—for example, a multi-party computation relay that requires several nodes to cooperate before revealing a block. Such designs are being explored by teams like Nethermind and Flashbots under the umbrella of “distributed relays.” Another trend is builder-specific relays that are run by builders themselves to bypass third-party fees. For instance, the Titan builder operates its own relay to capture both builder and relay profit, offering validators a unique integration.
Ultimately, the MEV-Boost relay protocol may be seen as a temporary bridge between the pre-PBS world of monolithic validators and a future of fully enshrined PBS. For now, mastering relay selection and understanding relay economics is essential for any validator aiming to maximize returns without sacrificing decentralization. Relays are not evil—they are simply the market-created solution to a protocol gap. As Ethereum evolves, the relay layer will adapt, but its core function as a blind auctioneer will likely persist in some form for years to come.
Common mistakes to avoid
- Relying on a single relay without fallback, risking missed proposals during outages.
- Ignoring relay fee structures and assuming the highest bid always yields the best net return.
- Not verifying relay censorship policies and accidentally supporting OFAC-compliant filters.
- Using the default mev-boost configuration without tuning relay URLs based on geographic latency.
- Assuming relays are neutral—some relays favor their own builder or charge hidden fees.
- Failing to monitor relay health via dashboards, leading to outdated or dead relay endpoints.
Frequently asked questions
How do MEV-Boost relays make money?
Most major relays (such as Flashbots, Ultra Sound, and Agnostic) run for free as public goods and do not take a cut of the bid. Some relay operators also run their own builders, capturing value on the block-building side rather than through relay fees.
Can a relay steal my block or MEV?
In theory yes, if the relay is malicious it could simulate your block and frontrun you. However, the blinded block design makes this difficult because the relay commits to the block hash before seeing the contents. Reputation and economic constraints prevent most relays from cheating.
What is the difference between Flashbots relay and Bloxroute relay?
Flashbots relay runs at no fee as a public good and historically applies OFAC filtering. Bloxroute operates multiple relays, including a max-profit relay (no OFAC filter) and regulated/ethical variants (OFAC filter). Bloxroute also generally has lower latency due to global infrastructure.
Do I need to run a relay as a builder?
No, builders only need to connect to existing relays via the builder API. However, large builders sometimes run their own relay to reduce fees and gain direct validator access, as Titan does.
Will ePBS replace MEV-Boost relays?
ePBS aims to encode the auction mechanism directly into the Ethereum protocol, removing the need for external relays. However, ePBS is still under research and not expected in the near future. Relays will remain essential in the interim.
How can I check if my relay is censoring?
You can inspect the relay’s source code or use block explorers that label relay origin. Community dashboards track which relays include transactions from sanctioned addresses and which apply OFAC filtering.
Related reading
Track the entities behind the concepts
DeFi Intel maps 11,000+ protocols, tokens and companies to a typed knowledge graph — with live data, incidents and regulation.