The House Edge

Appearance

6. The Infrastructure Layer ​

The firms you've never heard of that route most of the order flow — and what they sell, to whom, and why. ​

Cold open ​

If a Solana retail trader opens her Phantom wallet on a Tuesday afternoon in May 2026, signs a swap, and watches the transaction confirm within two seconds, she has interacted with — at minimum — Phantom (the wallet), Helius (the wallet's RPC backend), Jupiter (the aggregator), Jito Labs (the bundle market the aggregator routed through), and Jito-Solana (the validator client the leader who included her transaction is running). On the slot her transaction landed in, the leader may also have been running BAM — Jito's encrypted block-construction enclave — which means a sixth firm touched her trade between her click and its inclusion. Phantom is the only one whose name appears on her screen. The other five are infrastructure firms: companies whose business is to route, sequence, build, or refund trading flow on behalf of the firms the trader does see. This chapter is about who they are, what they sell, and why their existence is the most under-appreciated fact about on-chain trading.

What this chapter answers ​

  • Which firms operate the major layers of the on-chain trading infrastructure stack on Solana and on Ethereum, and what does each one actually sell?
  • Who buys at each layer, and what is the take rate?
  • Why have these firms emerged on top of "permissionless" chains that, in theory, no one needed to operate?
  • How big is this infrastructure layer in aggregate — what fraction of every on-chain trading dollar passes through it?

The setup ​

The on-chain trading infrastructure stack has four functional layers, each occupied by a small number of firms with measurable share at that layer. The first is the RPC provider: the API gateway that wallets and aggregators talk to when they want to read the chain's state or submit transactions to it. The second is the block builder or bundle market: the firm whose software assembles the actual contents of the next block from the available pool of transactions, bundles, and private order flow. The third is the relay (an Ethereum-specific layer that has no direct Solana equivalent): the entity that sits between block builders and validators, accepting bids from the former and forwarding them to the latter. The fourth is the Order Flow Auction (OFA) operator or refund service: the firm that captures a transaction's MEV potential at the wallet or RPC layer and either refunds a share back to the user or sells the right to fill it to a searcher.

Each of these layers, on every chain that supports any of them, is operated by a small number of firms. Solana's RPC market is dominated by a handful of providers (Helius being the largest single operator). Ethereum's block-building market is dominated by three firms (Titan, BuilderNet, Quasar account for ~91% of mainnet blocks). The relay market is similarly concentrated. The OFA market on Ethereum is split across a handful of named services (Flashbots Protect, MEV-Blocker now operated by Consensys's Special Mechanisms Group, CoW DAO's solver auction). The structural pattern: each function has its operators, the operators don't compete much across functions, and most operators occupy more than one layer.

This chapter is the third in Part II of the book — The Actors. Chapter 4 met the searcher; Chapter 5 met the validator and the builder. Chapter 6 meets the infrastructure firms behind those direct counterparties — the firms whose names the trader never sees but whose pipes the trader's transactions all flow through.

Two firms anchor the chapter. Helius is the cleanest single example on the Solana side: one firm that spans six layers of one chain's stack. Flashbots is the closest Ethereum equivalent, though its 2026 product stack is structurally different from its 2022 version. The other named firms — Jito Labs, Titan, BuilderNet, bloXroute, plus a long list of smaller and more specialised operators — get developed against those two anchors.

The worked example ​

Helius was founded in 2022 by Mert Mumtaz (CEO; very visible in Solana ecosystem commentary), Liam Vovk, and Nicolas Pennie. The firm raised a $9.5 million Series A in February 2024 led by Foundation Capital — Anatoly Yakovenko (Solana Labs co-founder) and Raj Gokal (Solana Labs co-founder) participated personally — and a $21.75 million Series B in September 2024 co-led by Haun Ventures and Founders Fund. PitchBook lists total raised at approximately $67.6 million.[1]

Helius operates at six distinct layers of the Solana stack simultaneously: as an RPC provider with paid subscription tiers from $49 per month (Developer) through $499 (Business), $999 (Professional), to bespoke enterprise contracts; as a streaming-data provider via LaserStream, the firm's gRPC ingestion product, included in the Business tier and above from 7 April 2026 onward; as a transaction-submission service via Sender, which routes signed transactions across seven regions and through Helius's own paid validator stake; as the network's largest single validator with over 15 million SOL staked at 99.9%+ slot success and 100% rewards passthrough to delegators; as a liquid-staking-token operator (hSOL, via the Sanctum launchpad); and as one of the four named launch operators of Jito's Block Assembly Marketplace (alongside Triton One, SOL Strategies, and Figment).[2] The firm's wallet customer roster includes Phantom, Backpack, Solflare, Squads, Bitwise (the exclusive Solana ETF staking partner), Bitgo, Ledger, Exodus, and Trust. Helius markets a 50/50 split on backrun-MEV rebates to the wallets that route transactions through it — the wallet's user captures half the extracted value, Helius keeps the other half.[3] The structural fact about the firm: most of the wallets the reader is likely to have heard of run on Helius's pipes.

Flashbots, the chapter's Ethereum anchor, looks structurally different. The firm was founded in 2020 by Phil Daian, Stephane Gosselin, Alex Obadia, Scott Bigelow, and Tina Zhen; headcount in March 2026 was approximately 55, with strategic direction led by Hasu and ongoing technical leadership by Phil Daian. Stephane Gosselin departed over disagreements about post-OFAC censorship choices and now leads Frontier Research and Primev. The firm's 2026 product stack is meaningfully different from its 2022 stack: Flashbots Protect (2.1 million+ Ethereum accounts; $43 billion+ in cumulative shielded DEX volume; 313 ETH in cumulative refunds returned to users); MEV-Share (the selective-disclosure orderflow auction protocol); Flashbots Relay (now ~2.15% of payload share on relayscan, down sharply from the firm's historical dominance); and a contribution to BuilderNet, the multi-operator block-building protocol Flashbots co-founded in November 2024 after stopping its centralised builder.[4] One product line was archived: SUAVE (the firm's ambitious cross-chain MEV-as-a-service blockchain) was wound down in May 2025, with the strategy refocused on BuilderNet and Flashnet (the network-anonymized mempool design Chapter 3 referenced).

The two firms occupy multiple layers of their respective stacks — RPC, OFA, validator/builder, refund routing — and that layer-spanning is the chapter's central pattern. The rest of the chapter describes the other firms at each layer, with Helius and Flashbots returning as comparison points throughout.

The mechanics, in detail ​

The RPC providers ​

Every transaction the chapter has discussed across five prior chapters has, at the bottom of its lifecycle, been broadcast to a chain through an RPC provider. The RPC is the API gateway between the user-facing application and the chain. The reader will rarely have interacted with an RPC provider directly — wallets, aggregators, and exchanges configure RPC endpoints on the user's behalf — but every transaction's outbound path runs through one.

On Solana, the RPC market in 2026 is dominated by a small number of operators. Helius is the largest by a wide margin: the firm's role as the RPC backend for the chain's most-used wallets (Phantom, Backpack, Solflare) and its paid-tier validator-stake-backed transaction-submission service give it a position that no competitor matches at scale. Triton One is the second major Solana RPC operator, with a multi-chain footprint (Solana, Sui, Monad, Pythnet) and a customer base that includes OpenBook, Pyth, and Wormhole; Triton was named in mid-2024 as the Solana Foundation's official partner on the RPC 2.0 initiative, an architectural overhaul of Solana's read layer.[5] Astralane is the newer entrant — a high-speed infrastructure firm whose product stack is explicitly aimed at on-chain market makers and HFT bots that need ultra-low-latency data and submission paths. The firm's a4.astralane.io indexing platform, Kafka streaming, and middleware are described in CoinDesk's February 2026 piece on Solana high-speed infrastructure as the HFT-tier alternative to Helius. Astralane's funding round and founders are not publicly disclosed; the firm is more opaque than its larger competitors.[6]

On Ethereum, the RPC market is structurally older and more fragmented than Solana's. The legacy multi-chain providers — Infura (Consensys), Alchemy, QuickNode — still serve most of the wallet and developer traffic, but specialised crypto-native firms (Flashbots Protect as a private-routing RPC; MEV-Blocker as an alternative private RPC; bloXroute as a multi-chain BDN provider) have taken share in the trade-flow segments of the market.

The structural observation that lands in the RPC subsection: the RPC layer is increasingly the surface at which MEV rebates are first captured. Helius's 50/50 split with its wallet customers is the canonical example — when a transaction routed through Helius creates a backrunning opportunity that Helius's validator partners capture, the wallet (Phantom or Backpack or Solflare) gets half the resulting MEV revenue as a refund flowing back through the pipeline, with the wallet free to share that with its user or keep it as platform revenue. The position is structurally similar to a US equity broker's payment-for-order-flow relationship with a wholesaler (Chapter 1 footnote 6) — except that the rebate flows in the opposite direction (PFOF flows from wholesaler to broker; the on-chain MEV rebate flows from RPC provider to wallet) and the value being captured is generated by extraction rather than execution-quality routing. The economic shape, however, is the same: an upstream party (the wholesaler in equities, the searcher in crypto) pays a downstream pipeline operator (the broker in equities, the RPC provider in crypto) for access to retail order flow.

Helius's customer concentration is the layer's most structurally interesting fact. Most of the major Solana wallets the reader has heard of share a single RPC backend. Phantom, Backpack, Solflare, Bitgo, Ledger, Exodus, Squads, Trust, and Bitwise (the exclusive Solana ETF staking partner) all run on Helius. The wallets compete on user-facing UX and brand; behind the interface, they share an infrastructure provider. The structural argument the chapter will land in the verdict: this is not unusual for infrastructure layers — most US equity brokers route to a small number of wholesalers — but it is a fact most retail users do not know.

The block builders and bundle markets ​

This subsection covers the firms whose software assembles the actual contents of the next block from the available pool of transactions, bundles, and private order flow. On Ethereum the layer is dominated by three named firms. On Solana the layer is structurally different — the validator constructs the block, but a small number of firms operate the infrastructure that shapes what the validator includes.

On Ethereum, the May 2026 builder landscape is the chapter's single most concentrated dimension. Live relayscan.io snapshots show Titan at approximately 51.55% of all mainnet blocks built in the 24-hour window ending 14 May 2026, BuilderNet at 24.09%, and Quasar at 15.29%. The top three builders account for approximately 91% of Ethereum's block-construction activity.[7] Each firm has a different operator structure and a different strategic position.

Titan is operated by Gattaca, a London-based crypto firm that evolved from a venture-backed trading shop into block-building infrastructure. Kubi Mensah is CEO and co-founder of Gattaca; the team is described in industry interviews as drawn from ex-miners, ex-validators, ex-searchers, and ex-traders across multiple chains. The firm has not publicly disclosed a seed or Series A funding round.[8] Titan operates two products: Titan Builder (the block-construction service) and Titan Relay (the firm's own relay), which carries approximately 26.65% of Ethereum's relay payload share — meaning Titan-the-firm is involved in well over half of Ethereum's mainnet blocks across its builder and relay businesses combined. Titan's reported builder margin under exclusive-flow arrangements (such as the deal with the Banana Gun trading bot) has reached approximately 17.75%, materially higher than Beaverbuild's pre-BuilderNet margin of approximately 9%.[9] The exclusive-flow business model is itself the seed for Chapter 7.

BuilderNet is the structural counterweight to Titan: a multi-operator builder protocol co-launched in November 2024 by Flashbots, Beaverbuild, and Nethermind. Each operator runs a node instance running in a Trusted Execution Environment; all operators share orderflow equally; an "open-source refund rule" redistributes excess builder profit back to orderflow originators based on contribution to winning blocks. Beaverbuild — once one of Ethereum's largest standalone builders — fully migrated its standalone operation into BuilderNet on 6 May 2025. BuilderNet's current share is approximately 24.09% of Ethereum builder market.[10] The protocol has been criticised, notably by EigenPhi, for permissioned onboarding: as of May 2026 the founding three operators remain the principal node operators, with no major new operator publicly added, prompting the charge that the protocol's decentralisation properties are partial.

Quasar is the chapter's most structurally interesting Ethereum builder. The firm holds approximately 15.29% of Ethereum's block-construction share — substantial enough to qualify as one of the three firms that, between them, build essentially every Ethereum block — but its operators are not publicly disclosed. The rumoured Manifold Trading connection is unconfirmed; the firm's founding team, funding history, and corporate structure are not in the public record. A firm running ~15% of Ethereum blocks with no publicly identifiable operators is itself a chapter-level point: the structural shape of the Ethereum block-building market is partially anonymous to the participants who fund it through their priority fees and MEV-Boost payments.

On Solana, the block-construction layer is operated by a different cast of firms because the architecture is different. The validator constructs the block, but Jito Labs operates the bundle market that supplies most of the high-value transactions and the BAM enclave that handles confidential block-construction for participating validators. Jito Labs was founded in 2021 and led by Lucas Bruder (CEO); the firm raised a $10 million Series A in August 2022 from Multicoin Capital and Framework Ventures. The JTO token launched in December 2023; market capitalisation as of 14 May 2026 was approximately $231 million at a price of $0.49.[11] The Jito product stack — Jito-Solana validator client, Block Engine, BAM, JitoSOL, the Jito DAO, the TipRouter — touches nearly every layer of Solana's MEV stack, and Jito-Solana client variants (the standard Jito, JitoBAM, and Jito within Frankendancer) collectively run approximately 94% of Solana stake by Q2 2025.

The chapter's load-bearing 2025 governance event lives here. JIP-24, passed by the Jito DAO in 2025, routes the entire 6% take on Block Engine and BAM tips to the Jito DAO treasury in perpetuity — replacing the prior 3%/3% split between Jito Labs (the company) and the DAO. Jito Labs no longer takes a direct cut from Block Engine or BAM fees. Jito Labs' historical revenue from these flows was substantial — approximately $25 million in Q4 2024 from Block Engine commissions alone, approximately $4.7 million in Q3 2025 — and projected annual run rate to the DAO post-JIP-24 is in the $15–50 million range.[12] The structural argument: a major infrastructure firm voluntarily restructured its revenue stream to route all the fees through a DAO. The chapter does not have another case this clean of an infrastructure firm doing this. JitoSOL (a 4% management fee on staking rewards across the largest Solana liquid staking token) remains the firm's primary direct revenue stream.

Two other Solana block-construction-adjacent firms warrant explicit mention. Temporal, the firm DL News reported in 2025 as the team behind the $40 billion-volume Solana "dark" AMM HumidiFi (Chapter 2), is led by Ben Coverston (ex-Citadel Securities, ex-Akuna, ex-MarginFi) with co-founder Kevin Pang (ex-Jump Trading, ex-Paradigm). Temporal also operates the Nozomi transaction-submission service (use.temporal.xyz/nozomi) and holds investments in DoubleZero, Backpack, and Sanctum.[13] Paladin is a Solana validator-side bot that runs inside the Jito client to prevent sandwiching and to extract atomic-arbitrage MEV for the validator, with a 90/5/5 reward split between the block leader, Palidators (the bot's validator participants), and PAL token holders. Paladin has been integrated by Chorus One and is one of the four named builders that Harmonic aggregates.[14]

The Solana block-construction layer has one structurally striking single-operator concentration fact: Ben Coverston is CEO of both Temporal AND Harmonic. Temporal operates HumidiFi (a prop-AMM that captures Solana spot DEX volume) and Nozomi (a transaction-submission service); Harmonic operates the open block-building marketplace at the validator-selection layer that Chapter 3 introduced. The two firms are structurally distinct businesses operating at different layers, but they share a CEO. The pattern — the same individual operating multiple firms at adjacent layers of one chain's infrastructure stack — is the kind of concentration the chapter's argument predicts but that no other chain has produced quite as visibly.

The relays and routing infrastructure ​

The Ethereum relay layer sits between block builders and validators. A relay accepts a builder's bid, holds the bid's full block contents in escrow, forwards the bid header to the proposer (the validator scheduled to propose the slot), waits for the proposer's signature on the header, then reveals the full block to the network. The structural function: the relay is a commitment device that prevents the proposer from extracting MEV by re-organising the builder's work after seeing it. The relay market in May 2026 is concentrated but not monopolised.

Live relayscan.io payload-share figures for the 24 hours ending 14 May 2026: Ultrasound Money at approximately 35.71%, Titan Relay at 26.65%, bloXroute Max-Profit at 13.44%, bloXroute Regulated at 11.83%, Aestus at 7.31%, Flashbots Relay at 2.15%.[15] The top three relay operators (Ultrasound, Titan, bloXroute) handle roughly three-quarters of payload share.

bloXroute Labs is the firm operating two of those relays. Founded in 2018 by Uri Klarman (CEO, Northwestern PhD), bloXroute is headquartered in Chicago with R&D in Israel. The firm has raised approximately $89.6 million across seven rounds, with a $70 million Series B in April 2022 led by SoftBank Vision Fund 2 — with Dragonfly, Flow Traders, GSR, Jane Street, Lightspeed, and ParaFi participating.[16] bloXroute is structurally unique among the named Ethereum relay operators in that it runs both a regulated (OFAC-filtered, US-compliant) relay and a neutral (profit-maximising) relay in the same product family. The combined Ethereum relay share is approximately 25.27%. bloXroute's broader product stack — the BDN (Blockchain Distribution Network, the firm's core fast-transaction-propagation product), private mempool services, MEV-defending RPC tooling — operates across Ethereum, BSC, Solana, and other networks; the firm is the closest the chapter has to a "multi-chain trading infrastructure" operator.

Aestus is a counterpoint to bloXroute's commercial scale. The relay is a non-profit operated by community contributors — one named operator is "KuDeTa" from the EthFinance / EthStaker community — and runs as a public good, serving over 650,000 validators as of February 2026.[17] Aestus has no firm-level revenue model and no funding history; it is the closest thing the chapter has to a "this is what an infrastructure layer looks like before commercial firms occupy it" point.

Flashbots Relay holds approximately 2.15% of relay payload share in May 2026. Flashbots' relay was historically dominant — at peak in 2023 it carried the majority of Ethereum's payload — and its 2% share in 2026 reflects the firm's strategic shift toward BuilderNet (where it operates as a co-founding builder) and away from running a centralised relay. Hasu's public framing of the layer's structural problem — "the absence of a neutral relay is a failure of the Ethereum ecosystem" (Chapter 3 footnote) — articulates Flashbots' position: the firm believes the structural concentration around Titan and bloXroute is itself a problem that the protocol needs to solve through ePBS, FOCIL, and the broader institutional roadmap Chapter 5 documented.

Solana has no analogous relay layer because the chain's matching engine is inside consensus and the validator constructs the block directly. The Helius and Triton "RPC + Sender + paid validator stake" stacks functionally compress what on Ethereum is two separate layers — RPC and relay — into a single product surface. The structural difference reflects each chain's underlying architecture: Solana's Gulf Stream + slot-leader-as-block-constructor model does not produce a separate relay role; Ethereum's MEV-Boost separation of proposer and builder explicitly requires one.

The Order Flow Auction operators and refund services ​

The fourth layer the chapter develops is the Order Flow Auction (OFA) operator — the firm that captures a transaction's MEV potential at the wallet or RPC layer, auctions the right to fill it to searchers or solvers, and refunds a share of the resulting bid back to the original submitter. The layer is most developed on Ethereum, where four named services compete: Flashbots Protect, MEV-Blocker (now operated by Consensys's Special Mechanisms Group), CoW DAO's solver auction, and 1inch Fusion (Chapter 3).

Flashbots Protect is the largest by user count: approximately 2.1 million Ethereum accounts have routed through it, with $43 billion+ in cumulative shielded DEX volume and 313 ETH in cumulative refunds returned to users. The mechanism: a user changes their wallet's RPC URL to the Flashbots Protect endpoint; transactions are then bundled with refunds from any backrunning value captured by searchers; the user's wallet still appears unchanged, but their transactions never enter the public mempool.[18] MEV-Share is Flashbots' selective-disclosure variant: a user can opt to share specific parts of their transaction (the path but not the amount, or vice versa) with searchers in exchange for a refund of the extracted value. Adoption has been steadier than Protect's but never matched the simpler URL-swap path Protect offers.

MEV-Blocker changed hands in January 2026. The service was originally operated by CoW DAO; it transferred to Consensys's Special Mechanisms Group (SMG), the Joseph Lubin-affiliated firm's blockchain-microstructure research division. SMG had been acquired by Consensys in 2024 as part of the firm's broader MEV-research expansion. MEV-Blocker at the time of acquisition had served over 4.5 million unique wallets and returned 6,177 ETH in cumulative rebates; the protocol returns approximately 90% of backrun surplus to users and 10% to validators.[19] CoW co-founder Anna George cited SMG's "technical depth and alignment with MEV Blocker's original mission" as the reason for the transfer.

CoW DAO itself operates a different product — the CoW Swap solver auction, which uses an intent-based architecture where the user signs an intent (rather than a specific transaction) and a pool of competing solvers bids to fill it. The protocol has processed over 73 million transactions and hundreds of billions of dollars in protected volume through February 2026. Revenue model: a 50% price-improvement share on limit and market orders, plus a 0.1% volume fee on the Gnosis Chain, Arbitrum, and Base deployments. CoW launched on Ink (Kraken's L2) on 17 February 2026.[20]

On Solana, the equivalent layer is operated by Helius (via the 50/50 backrun-MEV rebate split with wallet customers) and by Kolibrio, the VC-backed Order Flow Auction startup Chapter 3 introduced. Kolibrio runs the auction at the RPC layer: integrators (wallets, bots, solvers) route user transactions through Kolibrio, Kolibrio auctions the visibility to searchers, and the integrator receives a share of the rebates. Kolibrio raised a $2 million seed in February 2023 led by Jump Crypto.[21]

The structural pattern across the OFA layer: every operator captures value through a refund mechanism — the searcher pays, the user (or wallet, or integrator) receives a share back. The differences are in who captures what fraction and in which specific architectural primitive the operator uses (selective disclosure vs. full bundling vs. intent-based solving). The chapter's central observation: the OFA layer is the layer where on-chain markets most directly resemble TradFi's payment-for-order-flow ecosystem (Chapter 1), with the difference that the value flowing back to the retail user has been institutionally formalised as "MEV refund" rather than absorbed into the venue's spread.

Meet the Infrastructure Provider

Job: Operate one or more layers of the on-chain trading infrastructure stack — RPC, OFA, builder, relay, block engine — and capture a share of the trading flow that passes through.

How they earn: Subscription tiers (Helius's $49–$999+ per month for RPC and streaming); take rates on bundled transactions (Jito's pre-JIP-24 6% Block Engine split, BAM's continuing 6% to the DAO); builder margins (Titan's reported 17.75% under exclusive flow; BuilderNet's equal-split open-refund-rule model); relay-side commissions; MEV-rebate flows (Helius's 50/50 with wallets, MEV-Blocker's 90/10 to users). The take is layered: a single trade can pay an RPC fee, an OFA share, a builder margin, a relay commission, and a validator MEV-Boost payment, each going to a different firm.

How they spend: Low-latency hardware and bandwidth (the dominant cost for builders, relays, and Solana RPC operators); engineering teams (small — Helius's RPC operation, Jito's full stack, and Titan's builder are each operated by teams of dozens, not hundreds); validator stake (the firms that operate validators also lock up substantial capital — Helius's 15M+ SOL is a meaningful balance sheet item); and infrastructure partnerships (BAM Node operator costs at Jito, relay-fee splits at bloXroute).

The moat: Customer relationships and integration depth. Helius is hard to dislodge as Phantom's RPC backend because the integration has been deep for years and the rebate flow is now part of Phantom's revenue model. Titan is hard to dislodge as the dominant Ethereum builder because the firm has built private-flow relationships with specific trading firms (Banana Gun being the named example) that BuilderNet cannot replicate without their consent. The moat is not the technology — most of these firms' core software is replicable — but the customer base and the flow relationships.

TradFi analogue: The B2B infrastructure that sits behind every modern equity exchange — DTCC for clearing, NYSE's matching engine for the on-exchange tape, the wholesalers (Citadel Securities, Virtu) that internalise retail flow before it ever reaches the exchange, the technology vendors (Refinitiv, Bloomberg) that provide the data feeds. Retail traders interact with brokers; the brokers interact with this infrastructure. On-chain, the equivalent ecosystem emerged in roughly four years (2022–2026) versus four decades in equities.

Founded: Various — Flashbots in 2020, bloXroute in 2018, Jito Labs in 2021, Helius in 2022, BuilderNet in 2024, Astralane and Harmonic later still. The dates compress the speed at which this layer was built.

How this plays out on each chain ​

On Solana, the infrastructure stack is most striking for the concentration of operators across layers. Helius spans six layers (RPC, streaming, transaction submission, validator stake, LST, BAM Node). Jito Labs spans the validator-client, Block Engine, and BAM layers, with JitoSOL adding LST exposure. Temporal operates HumidiFi (prop AMM) and Nozomi (transaction submission), with its CEO Ben Coverston also operating Harmonic (the block-building marketplace) at a different layer. Jump Crypto maintains Frankendancer and Firedancer (the validator-client business) and led Kolibrio's seed (the OFA layer). Anza maintains Agave (the reference client) with a 13% Solana Labs stake interest. The single-firm-spans-multiple-layers pattern is the chain's structural shape.

On Hyperliquid, there is no equivalent third-party infrastructure layer. The chain runs its own validator set, its own matching engine, and its own HLP vault — the surfaces that on Solana and Ethereum have produced dozens of third-party firms have, on Hyperliquid, been internalised into the protocol's permissioned-top-21 validator set. The infrastructure firms in this chapter do not operate on Hyperliquid because the architecture does not need them. Chapter 9 develops what that means for the chain's competitive position.

On Ethereum and its L2s, the infrastructure layer is broader, more functionally specialised, and less individually concentrated than Solana's, but with sharper concentration at specific layers (Titan's 51% of blocks; bloXroute's 25% of relay share). Flashbots at the OFA and relay layers, Titan / Gattaca at the builder and relay layers, BuilderNet as the multi-operator builder alternative, bloXroute spanning relay and BDN, Consensys / SMG at the OFA layer, CoW DAO at the solver-auction layer. The L2 sequencer firms (Coinbase Cloud at Base, Offchain Labs at Arbitrum, the Optimism Foundation's chosen sequencer for OP Mainnet) operate at a structurally similar position to "builder + relay + sequencer combined," but Chapter 10 develops them in that frame.

The structural argument the comparison lands: the Solana infrastructure layer is concentrated by operator (the same firms appearing at multiple layers); the Ethereum infrastructure layer is concentrated by layer (different firms but a small number of operators at each specific function).

Who wins, who loses, why ​

Winners. The infrastructure firms themselves. Helius — six layers of the Solana stack, $67.6 million raised, the RPC pipeline behind essentially every named Solana wallet, the network's largest single validator, an exclusive Solana ETF partnership with Bitwise. Jito Labs / Jito Foundation — the Block Engine, BAM, JitoSOL, the Jito DAO; post-JIP-24 the entire 6% Block Engine and BAM tip take routes to the DAO treasury rather than to the firm, but the company that built the products continues to operate them and to capture JitoSOL's 4% management fee. Titan / Gattaca — 51.55% of Ethereum blocks; Titan Relay's 26.65% of payload share; reported margins to 17.75% under exclusive flow arrangements. bloXroute — $89.6 million in total funding; ~25% combined Ethereum relay share; the only major operator running both a regulated and a neutral relay. Flashbots — 2.1M+ Protect users; co-operating BuilderNet; the firm whose strategic pivot from SUAVE to BuilderNet/Flashnet defines the Ethereum infrastructure landscape's institutional response to centralisation. And smaller specialist firms across both chains: Triton, Astralane, Kolibrio, Paladin, Aestus, the Consensys/SMG team that absorbed MEV-Blocker.

The chapter's full firm-by-layer landscape, consolidated as a single reference table:

FirmChain(s)Founder / CEOFoundedFundingPrimary layer(s)Current scale
HeliusSolanaMert Mumtaz2022~$67.6M (Series A + B)RPC · streaming · transaction submission · validator · LST · BAM Node15M+ SOL staked (largest single validator); RPC backend for Phantom, Backpack, Solflare, Bitwise SOL ETF
Jito Labs / FoundationSolanaLucas Bruder2021$10M Series ABlock Engine · BAM · validator client · LSTJito-Solana variants ~94% of Solana stake; JTO market cap ~$231M; post-JIP-24 all 6% Block Engine + BAM tips to DAO
Triton OneSolana (+ Sui · Monad · Pythnet)not publicly namedn/an/aRPC · BAM NodeSolana Foundation RPC 2.0 partner; OpenBook, Pyth, Wormhole customers
TemporalSolanaBen Coverstonn/an/aProp AMM (HumidiFi) · transaction submission (Nozomi)HumidiFi ~$40B cumulative volume; Nozomi competitor to Helius Sender
HarmonicSolanaBen Coverston · Jakob Povšič2025$6M Paradigm-led seedBlock-building marketplace~17% of Solana stake; auctions Jito, Temporal, JitoBAM, Paladin
AnzaSolanaStephen Akridge · Jed Halfon2024SOL Foundation grantsValidator client (Agave)Solana Labs retains 13% stake; pure protocol engineering
Jump CryptoSolana (multi-chain)Jump Trading Group2018self-fundedValidator client (Firedancer · Frankendancer)Frankendancer 12% of stake; led Kolibrio seed
RakuraiSolanaAli Rizvi2024$3M Anagram-led seedValidator client (closed-source Agave fork)~6% of Solana stake; Figment migration anchor
KolibrioSolana (+ Ethereum)Alex Starikov · Anatolii Padenko2022$2M Jump-led seedOFA at RPC layerLisbon-based; "Meow RPC"
PaladinSolanacommunity / token-governedn/an/aValidator-side anti-sandwich botReward split 90/5/5 leader/Palidators/PAL holders
AstralaneSolananot publicly disclosedn/anot publicly disclosedHFT-tier RPC · indexing · streaminga4.astralane.io platform; CoinDesk Feb 2026
FlashbotsEthereum (+ chains via BuilderNet)Phil Daian · Hasu (strategy)2020Paradigm-led seedOFA (Protect, MEV-Share) · relay · BuilderNet co-operator2.1M+ Protect users; $43B+ shielded; ~55 headcount
Titan / GattacaEthereum (+ Arbitrum via Timeboost)Kubi Mensah~2023 (Gattaca origins earlier)not publicly disclosedBuilder · relay~51.55% of Ethereum builder share; ~26.65% Titan Relay share
BuilderNetEthereumco-operated (Flashbots · Beaverbuild · Nethermind)November 2024jointly operatedMulti-operator builder~24.09% of Ethereum builder share; open-source refund rule
bloXrouteEthereum · BSC · Solana · multi-chainUri Klarman2018~$89.6M (Series B led by SoftBank Vision Fund 2)Two relays (Max-Profit + Regulated) · BDN · private mempool~25.27% combined Ethereum relay share
QuasarEthereumnot publicly disclosedn/anot publicly disclosedBuilder~15.29% of Ethereum builder share
AestusEthereumcommunity (KuDeTa named)n/anon-profitRelay~7.31% relay share; 650K+ validators
Special Mechanisms Group / ConsensysEthereumJoseph Lubin (parent firm)acquired by Consensys 2024Consensys-fundedOFA (acquired MEV-Blocker January 2026)4.5M+ MEV Blocker users at acquisition; 6,177 ETH returned
CoW DAOEthereum · Gnosis · Arbitrum · Base · InkAnna George2020DAO-fundedSolver auction · intent-based OFA73M+ cumulative transactions; 50% price-improvement share

A directional aggregate, assembled bottom-up because no single source publishes the number: Solana's 2025 total network revenue was approximately $1.4 billion, of which approximately $720 million was MEV capture. A meaningful fraction of that flows through the firms named above. Jito Block Engine alone produced approximately $25 million of revenue in Q4 2024; the projected post-JIP-24 annual run rate to the DAO is $15–50 million. Ethereum aggregate MEV-Boost payments cumulative since the Merge are approximately $300 million. Helius's tiered subscription business plus validator stake plus rebate flows is opaque but substantial. The combined infrastructure-layer take across both chains in 2026 is in the mid-hundreds-of-millions to low-billions of US dollars in annual run rate — the precise number is not in the public record, but the order of magnitude is. The single most striking concentration fact, again: one firm (Titan/Gattaca) is involved in approximately 78% of Ethereum mainnet blocks across its builder (~51%) and relay (~27%) businesses combined.

Losers. The firms that did not survive the consolidation are the clearest cases. Eden Network wound down on 12 August 2025, citing the "fiercely competitive and costly" MEV relay/builder space; the firm's treasury (2,000 ETH) was distributed to EDEN holders at 0.00001506 ETH per token through 30 September 2025.[22] Beaverbuild retired its standalone block-building operation on 6 May 2025 and now operates as a BuilderNet operator only. Beyond firm-level losses, the indirect losers are the retail traders whose transactions pay a cumulative cut at every layer — RPC fee, OFA share, builder margin, relay commission, validator MEV-Boost payment, each routed to a different firm — and the validators without infrastructure relationships, whose ability to compete is constrained by which builders and relays they can access (Chapter 5).

Is this bad? The clinical answer is that infrastructure firms emerged because there were real operational gaps on permissionless chains. Running RPC at the scale Phantom requires is expensive; constructing optimal blocks under MEV-Boost is a hard engineering problem; refunding MEV requires sophisticated routing logic and trust relationships. The firms that filled the gaps captured the resulting margin. The structural concern is the concentration of that captured margin: Helius's six-layer Solana presence is the cleanest single case, but Titan's 51%-of-Ethereum-blocks dominance, Ben Coverston's CEO-of-two-firms position, and Quasar's anonymous 15%-builder share are all products of the same dynamic. The retail trader's question — who am I trading against, and who is taking a cut of every transaction I sign? — has an answer that the trader cannot easily discover from any user interface and that, even when developed firm-by-firm as this chapter has done, points to a small number of named operators capturing a substantial share of every on-chain trading dollar. That structure is not, by itself, evidence of misconduct. It is evidence that the chains' formal openness has been operationally enclosed by a B2B layer the average user does not know exists.

What changes when… ​

What changes when one of these infrastructure firms not only spans multiple layers but explicitly contracts with one specific searcher or trading firm for exclusive flow at one of those layers? When the take rate at the layer becomes a captive economic rent rather than a market-clearing price? Titan's reported Banana Gun arrangement is the seed; the structural answer is Chapter 7.

Footnotes and sources ​

  1. Helius — Funding Announcement, https://www.helius.dev/blog/funding-announcement. Series A $9.5M (February 2024, Foundation Capital, with personal participation from Anatoly Yakovenko and Raj Gokal); Series B $21.75M (September 2024, co-led by Haun Ventures and Founders Fund). PitchBook total raised approximately $67.6M. Founders: Mert Mumtaz (CEO), Liam Vovk, Nicolas Pennie. Accessed 2026-05-14. ↩︎

  2. Helius — Pricing, https://www.helius.dev/pricing; Solana Compass — Helius Validator, https://solanacompass.com/staking/helius; Helius — Block Assembly Marketplace (BAM), https://www.helius.dev/blog/block-assembly-marketplace-bam. Pricing tiers: Free $0 / Developer $49/mo / Business $499/mo (includes LaserStream from 7 April 2026) / Professional $999/mo / Enterprise custom. Validator stake: >15M SOL (largest single Solana validator), 99.9%+ slot success, 100% rewards passthrough, 0% commission. BAM launch operators: Helius, Triton One, SOL Strategies, Figment. Accessed 2026-05-14. ↩︎

  3. Helius — Backrun Rebates Documentation, https://www.helius.dev/docs/sending-transactions/backrun-rebates. The 50/50 split between Helius and the integrator (wallet) is the canonical 2026 RPC-layer MEV-refund model. Wallet customer list per Helius's Wallets Use Case page, https://www.helius.dev/use-case/wallets, and the Bitwise SOL ETF partnership at https://www.helius.dev/blog/bitwise-solana-etf. Accessed 2026-05-14. ↩︎

  4. Flashbots — Flashbots Protect Documentation, https://docs.flashbots.net/flashbots-protect/overview; Flashbots — Migrating to BuilderNet, https://writings.flashbots.net/migrating-to-buildernet; Tracxn — Flashbots company profile, https://tracxn.com/d/companies/flashbots/. SUAVE archival in May 2025 documented in the Flashbots writings archive. Headcount approximately 55 in March 2026 per Tracxn. Phil Daian co-founder; Hasu strategy lead; Stephane Gosselin departed for Frontier Research / Primev. Accessed 2026-05-14. ↩︎

  5. Triton — Announcing RPC 2.0 with Solana Foundation, https://blog.triton.one/announcing-rpc-2-0-with-solana-foundation-rethinking-solanas-read-layer-from-the-ground-up/. Triton's multi-chain footprint (Solana, Sui, Monad, Pythnet) and customer base (OpenBook, Pyth, Wormhole) documented on triton.one. Accessed 2026-05-14. ↩︎

  6. Astralane corporate site, https://astralane.io/; CoinDesk — Solana High-Speed Infrastructure, 23 February 2026, https://www.coindesk.com/business/2026/02/23/solana-company-starts-building-high-speed-infrastructure-to-prepare-sol-for-next-super-cycle. The firm's a4.astralane.io indexing platform plus Kafka streaming plus middleware position it as the HFT-tier Solana RPC alternative to Helius. Founding team and funding round size are not publicly disclosed. Accessed 2026-05-14. ↩︎

  7. relayscan.io, MEV-Boost Relay and Builder Statistics (24-hour snapshot ending 14 May 2026), https://www.relayscan.io/. Builder shares: Titan 51.55%, BuilderNet 24.09%, Quasar 15.29%, Beaverbuild 1.83%. Live data; will drift. Accessed 2026-05-14. ↩︎

  8. Scraping Bits — Kubi Mensah: Titan's Takeover podcast, https://rss.com/podcasts/scrapingbits/1090708/; titanbuilder.xyz corporate site. No public seed or Series A funding round disclosed for Gattaca/Titan. A separately-named "Titan" Solana meta-DEX aggregator raised $7M from Galaxy Ventures in September 2025 — this is an unrelated entity with the same name. Accessed 2026-05-14. ↩︎

  9. Observers — How Two Block Builders Monopolized Ethereum Block Production, https://www.observers.com/how-two-block-builders-monopolized-ethereum-block-production/, citing Titan disclosures around the Banana Gun arrangement. Already cited in Chapter 5 footnote 5. Accessed 2026-05-14. ↩︎

  10. BuilderNet — Introducing BuilderNet, https://buildernet.org/blog/introducing-buildernet; BuilderNet — Beaverbuild Migration, https://buildernet.org/blog/beaverbuild. Founding operators: Flashbots, Beaverbuild, Nethermind (November 2024 launch). Beaverbuild fully migrated its standalone builder operation into BuilderNet on 6 May 2025. EigenPhi's critique at BuilderNet Infrastructure Monoculture, https://eigenphi.substack.com/p/guest-post-buildernet-infra-monoculture. Accessed 2026-05-14. ↩︎

  11. Jito Labs corporate background; CoinMarketCap / CoinGecko for JTO market cap and circulation (14 May 2026). Series A $10M August 2022, Multicoin Capital and Framework Ventures. Accessed 2026-05-14. ↩︎

  12. Jito Forum — JIP-24: Jito DAO Receives All Block Engine Fees and Future BAM Fees, https://forum.jito.network/t/jip-24-jito-dao-receives-all-jito-block-engine-fees-and-future-bam-fees/860; Jito Network — September 2025 Monthly Roundup, https://www.jito.network/blog/september-monthly-roundup-2025/; CoinDesk — Jito Launches BAM, https://www.coindesk.com/tech/2025/07/21/jito-launches-bam-to-reshape-solanas-blockspace-economy. JIP-24 passed 2025; pre-JIP-24 the 6% take was split 3%/3% between Jito Labs and the Jito DAO. Q4 2024 Jito Labs revenue from Block Engine commissions approximately $25M; Q3 2025 approximately $4.7M; projected post-JIP-24 annual run rate to DAO $15-50M. JitoSOL 4% management fee remains. Accessed 2026-05-14. ↩︎

  13. DL News — Temporal Said to Be Behind Solana Prop AMM HumidiFi, https://www.dlnews.com/articles/defi/temporal-said-to-be-behind-solana-prop-amm-humidifi/; Temporal corporate background per industry interviews. Ben Coverston CEO (ex-Citadel Securities, ex-Akuna, ex-MarginFi); Kevin Pang co-founder (ex-Jump Trading, ex-Paradigm). Temporal operates Nozomi at https://use.temporal.xyz/nozomi/transaction-submission. Portfolio investments in DoubleZero, Backpack, Sanctum. Accessed 2026-05-14. ↩︎

  14. Paladin Documentation, https://docs.paladin.one/; Chorus One — Integration with Paladin, https://chorus.one/articles/chorus-ones-integration-with-paladin-reshaping-mev-on-solana. Reward split: 90% to block leader, 5% to Palidators + stakers, 5% to PAL token holders. One of four named builders Harmonic aggregates (Jito, Temporal, JitoBAM, Paladin). Accessed 2026-05-14. ↩︎

  15. relayscan.io (op. cit. footnote 7). Relay shares: Ultrasound Money 35.71%, Titan Relay 26.65%, bloXroute Max-Profit 13.44%, bloXroute Regulated 11.83%, Aestus 7.31%, Flashbots Relay 2.15%. Accessed 2026-05-14. ↩︎

  16. bloXroute corporate site, https://bloxroute.com/; Tracxn bloXroute profile, https://tracxn.com/d/companies/bloxroute/. Founded 2018; CEO Uri Klarman (Northwestern PhD); HQ Chicago, R&D Israel. Funding total approximately $89.6M across seven rounds; $70M Series B April 2022 led by SoftBank Vision Fund 2 (with Dragonfly, Flow Traders, GSR, Jane Street, Lightspeed, ParaFi). Accessed 2026-05-14. ↩︎

  17. Aestus — corporate site, https://aestus.live/; Flashbots Collective — Aestus: A Neutral Relay, https://collective.flashbots.net/t/aestus-a-neutral-relay/786. Run by community contributors (one named: "KuDeTa") from EthFinance / EthStaker. Non-profit, no firm-level revenue model; over 650,000 validators served by February 2026. Accessed 2026-05-14. ↩︎

  18. Flashbots Protect documentation (op. cit. footnote 4). 2.1M+ accounts; $43B+ cumulative shielded DEX volume; 313 ETH cumulative refunds returned to users. Accessed 2026-05-14. ↩︎

  19. Consensys — Consensys Acquires SMG to Deliver State-of-the-Art Solutions, https://consensys.io/blog/consensys-acquires-smg-to-deliver-state-of-the-art-solutions; CoW DAO — Special Mechanisms Group Acquires MEV Blocker RPC, https://cow.fi/learn/special-mechanisms-group-acquires-mev-blocker-rpc-to-advance-state-of-the-art-backrunning-auction-infrastructure. SMG acquisition by Consensys in 2024. MEV Blocker transferred from CoW DAO to SMG in January 2026. MEV Blocker at time of acquisition: 4.5M+ unique wallets, 6,177 ETH cumulative rebates returned. Protocol returns approximately 90% of backrun surplus to users, 10% to validators. Accessed 2026-05-14. ↩︎

  20. CoW DAO — Monthly Recap February 2026, https://cow.fi/learn/cow-dao-monthly-recap-february-2026. 73M+ transactions; hundreds of billions in protected volume cumulative. Revenue model: 50% price-improvement share on limit and market orders plus 0.1% volume fee on Gnosis Chain / Arbitrum / Base. Launched on Ink (Kraken L2) 17 February 2026. Accessed 2026-05-14. ↩︎

  21. Kolibrio Documentation, https://docs.kolibrio.xyz/; AIN.ua — Ukrainian crypto startup Kolibrio raises $2M, https://en.ain.ua/2023/02/16/ukrainian-crypto-startup-kolibrio-raises-2m-to-build-web3-service/. Already cited in Chapter 3 footnote 26. Accessed 2026-05-14. ↩︎

  22. The Block — Eden Network Shuts Down, https://www.theblock.co/post/366711/eden-network-shuts-down. Wound down 12 August 2025. Treasury (2,000 ETH) distributed to EDEN token holders at 0.00001506 ETH per EDEN through 30 September 2025. Eden cited the "fiercely competitive and costly" MEV relay/builder space. The chapter mentions Eden as the cleanest example of the firms that did not survive the 2024–2026 consolidation. Accessed 2026-05-14. ↩︎