The House Edge

Appearance

Chapter 10 — Research Note (Phase 1) ​

Status: RESEARCH (Phase 1 — written by agent, awaiting Nick's review) Date: 2026-05-14 Working chapter: 10 — Ethereum and Its L2s: The Most Studied, Least Settled Part: III — The Architectures (Ch 8 Solana done; Ch 9 Hyperliquid done; Ch 10 closes Part III)

How to read this note ​

Ch 10 closes Part III's three-chain architecture sequence. Its job is to land Ethereum as the third structural option in the book's chain-by-chain comparison: the chain where the extraction surface is most explicitly architected (PBS), most extensively measured (Wu et al., Pahari-Canidio, mevboost.pics, relayscan.io), and most concentrated by named operators (Titan 52%, BuilderNet 25%, Quasar 15%; Lido 24%, Coinbase 5%; Base capturing ~62% of L2 fee revenue). Solana compresses the extraction surface (Ch 8). Hyperliquid eliminates it (Ch 9). Ethereum architects it — and the architecture has been studied, measured, and partially regulated in ways the other two chains have not.

Seven things to know up front before drafting:

  1. The Ethereum side is largely already developed in Chs 5 and 7. Wu et al.'s 75 EOF arrangements / ~71% of trading-related builder revenue (Ch 7), Pahari-Canidio's 77.2%–84% fee share for exclusive transactions (Ch 7), the Titan/Banana Gun trace (Ch 7), the Titan 51.5% / BuilderNet 24% / Quasar 15.3% concentration (Ch 5), Coinbase Cloud's 4.5M ETH / 12.17% network share (Ch 5), and the Pectra / EIP-7251 max-effective-balance change to 2,048 ETH (Ch 5) are all in the book. Ch 10 must reference, not re-litigate, this material. The chapter's new contribution is the L2 layer, cross-chain MEV, restaking, and the integration of the three (PBS + L2 fragmentation + restaking as the chain's structural shape in 2026).

  2. The L2 landscape in May 2026 is concentrated and consolidating fast. Three rollups — Arbitrum, Base, Optimism — process approximately 90% of all L2 transaction volume. Total L2 TVL is approximately $48 billion across ~73 active rollups, with Arbitrum at ~$14.9–16.9B (40–44% share), Base at ~$10.7–11.2B (~28–33%), OP Mainnet at ~$5.6B, zkSync Era at ~$4.1B, Linea at ~$3.4B, Scroll at ~$2.1B, Starknet at ~$1.5B. (L2BEAT TVS; Spoted Crypto L2 Comparison 2026; accessed 2026-05-14)

  3. Base captures the L2 fee economy. The chapter's load-bearing L2 stat. Base captures approximately 62% of all L2 fee revenue, 46.6% of L2 DeFi TVL, and 70% of L2 active addresses. Base's daily sequencer revenue averages ~$185,291; priority fees alone average ~$156,138/day and constitute ~86.1% of daily sequencer revenue. ~64.9% of all priority fees on Base came from just 250 addresses in one year. Coinbase reported $109.4M of on-chain revenue in Q1 2026, with Base generating ~$870M lifetime revenue and the only L2 that was profitable in 2025 (~$55M net). (growthepie; CoinLaw L2 Gas Fee Statistics 2026; Pine Analytics — The Compression of L1 Value Capture; Yellow Research — Ethereum L2 Fee Revenue Competition; accessed 2026-05-14)

  4. The L1 fee revenue has collapsed; the L2 fee revenue partially replaced it but not at parity, and the L2-to-L1 payment has collapsed by ~90% post-Pectra/Fusaka. In 2024, L2s paid Ethereum ~$113M; in 2025, L2s paid Ethereum only ~$10M (a >90% decline). Ethereum's L1 daily gas revenue fell from a peak of ~$23M/day to ~$6.3M/day in 2025; Ethereum is now mildly inflationary (~0.23% annual) post-Pectra. The "ultrasound money" framing is, as of 2026, contested. The chapter must frame this as the L2 architecture as a value-capture problem for L1, not just as a scaling solution. (Pine Analytics; Bitcoin Ethereum News — Fusaka raises blob fees; accessed 2026-05-14)

  5. Cross-chain MEV is now a measurable economic surface, not a forward-looking one. The Cross-Chain Arbitrage paper (Flashbots + TU Munich, ACM SIGMETRICS 2024; arXiv 2501.17335) identified 242,535 cross-chain arbitrages across nine blockchains over Sept 2023–Aug 2024, totaling ~$868.64M trading volume and generating ~$10.05M searcher revenue / $8.65M net profit. 58.35% of activity was L1↔L2 pairs; 35.67% L2↔L2. Most trades use pre-positioned inventory (66.96%, settling in ~9 seconds); bridge-based arbitrages take ~242 seconds. Activity grew 5.5× over the study period and surged after Dencun (March 2024). Top five addresses execute >50% of all trades; one alone captures ~40% of daily post-Dencun volume. This is the chapter's "L2 fragmentation creates a new extraction surface" anchor. (Flashbots / Maire et al. — arXiv 2501.17335; Flashbots Collective forum; accessed 2026-05-14)

  6. EigenLayer's slashing went live (April 17, 2025); TVL is ~$15.3–19B at ~1,900 active operators; AVS roster is real but revenue is modest. EigenLayer commands ~94% of restaking market share with ~$15.258B TVL / 4.36M ETH (early 2026 figure) or ~$19B / 4.6M ETH (per other sources) or recently $25B per Mitosis aggregation. AVS revenue to operators remains small relative to TVL. Slashing is live but rarely invoked; redistributable slashing went mainnet via ELIP-006 with funds either burned (standard sets) or redistributed (redistributable sets). The chapter's relevance: restaking creates an additional cross-chain extraction surface (an Ethereum-staked validator now secures arbitrary AVSs, some of which operate cross-chain). (BlockEden — EigenLayer Crosses $18B; Mitosis University — Restaking Economy Hits $25B; Blockdaemon — EigenLayer Mainnet Slashing; accessed 2026-05-14)

  7. Glamsterdam / ePBS is delayed; the ePBS slip Ch 5 documented continues. EF Checkpoint #9 (April 2026) confirmed Glamsterdam Q2 2026 is unlikely; ePBS implementation is "trickier than anticipated." The Justin Drake "Strawmap" (Feb 2026) outlines a seven-fork roadmap through 2029 with one consensus + one execution headliner per fork. Glamsterdam current target is May/June 2026 but realistic slip is Q3/Q4 2026 with risk of further delay (Base team has warned the addition of FOCIL could push into 2027). The chapter must frame ePBS as "the protocol-level fix that has not arrived, while the MEV-Boost/relay architecture remains the de facto PBS." (EF Checkpoint #9; Blockonomi — Justin Drake's Strawmap; accessed 2026-05-14)

1. Key claims ​

Each numbered claim is something the chapter is allowed to state. Sources cited inline. The chapter's load-bearing claims are 5 (the L2 fee economy concentrates on Base), 8 (cross-chain MEV is a measured surface), and 14 (the PBS + L2 fragmentation + restaking trifecta).

Ethereum L1 in 2026 — the staked-but-shrinking layer

  1. Ethereum's validator count in May 2026 is approximately 1.06–1.1 million; total staked ETH is approximately 35.86 million, representing ~28.91% of total supply. Up from 32M ETH in early 2025 (29%) but with validator count gradually compressing as Pectra-era consolidations replace 32-ETH validators with up-to-2,048-ETH consolidated ones. (CoinLaw — ETH Staking Statistics 2026; Datawallet — Ethereum Staking Statistics 2026; beaconcha.in; accessed 2026-05-14)

  2. The Pectra upgrade activated 7 May 2025 (Epoch 364032) and raised the maximum effective balance per validator from 32 ETH to 2,048 ETH via EIP-7251, reducing network message overhead by ~40% and enabling consolidation of large operator stakes into fewer, larger validators. Already developed Ch 5; Ch 10 must reference but not re-litigate. (CoinDesk — Ethereum Activates Pectra; Consensys — Pectra overview; accessed 2026-05-14)

  3. Ethereum L1 fee revenue has collapsed since the EIP-4844 / Dencun activation (March 2024) and the post-Pectra L2 efficiency gains. Daily L1 gas revenue fell from a peak of ~$23M/day to ~$6.3M/day in 2025. ETH is now slightly inflationary (~0.23% annually). Average daily ETH burn fell to ~3.26 ETH/day post-Pectra (a ~71% decrease). Industry estimates suggest blob fees could contribute 30–50% of total ETH burn by 2026 depending on L2 activity. Ch 10 must frame this as a deliberate trade-off: the L2 architecture moved the fee economy off L1, and L1 has not (yet) recovered the value capture. (CoinLaw — Ethereum Gas Fees 2026; Bitcoin Ethereum News — Fusaka raises blob fees; accessed 2026-05-14)

  4. Ethereum's REV rank has slipped to #3 behind Hyperliquid and Solana by Q1 2026. Hyperliquid Q1 2026 app revenue ~$144.8M (#1 crypto app); Solana Q1 2026 chain REV ~$89.5M (#2); Ethereum Q1 2026 chain fees ~$82M but with much higher proportional capture going to L2s rather than L1. Hyperliquid passed Ethereum first by application metric, then Solana for several weeks (Ch 8). The chapter must frame Ethereum's structural position as the chain whose architecture intentionally moved value off L1, where Solana's and Hyperliquid's architectures kept value on L1. (Coinpedia on X — Q1 2026 ranking; PANews — Solana Q1 2026 Report; Odaily — L1 Value Capture Shrinks; accessed 2026-05-14)

PBS / MEV-Boost in 2026

  1. Live MEV-Boost telemetry for early 2026 (relayscan.io 24-hour snapshots at draft date):

    • Builder shares: Titan ~52.16%, BuilderNet ~24.63%, Quasar ~15.06% (Quasar's share has compressed slightly from 15.3% in May 2025), Eureka ~2.41% (the new entrant), Beaverbuild ~1.81%, Bombora ~1.31%. Top three ~92%.
    • Relay shares: Ultrasound ~35–36%, Titan Relay ~26%, BloXroute Max-Profit ~13%, BloXroute Regulated ~12%, Aestus ~7%, Flashbots Relay ~2%.
    • Top three builders + top four relays = ~92% / ~86% concentration; the structural shape is largely unchanged from Ch 5's framing but Eureka is new: a "programmable blocks" builder that raised $6.7M in 2025 from Spark Capital + Collider Ventures, founded December 2024, now fourth-largest by volume at ~1.5% market share. (relayscan.io; Wahrstätter on X — MEV-Boost Update; The Block — Eureka Labs raises $6.7M; accessed 2026-05-14)

  2. ePBS (EIP-7732) has slipped past Glamsterdam's original H1 2026 target. Per the EF's April 2026 Checkpoint #9 publication: "Glamsterdam in Q2 seems to me to be unlikely." The implementation is "trickier than anticipated" due to two-party consensus coordination and "partial block" handling. The first generalized Glamsterdam devnet was targeted for the week after Checkpoint #9. Justin Drake's "Strawmap" (Feb 2026) outlines a seven-fork sequence through 2029 with explicit pacing: one consensus headliner + one execution headliner per fork. Glamsterdam realistic mainnet activation: Q3/Q4 2026, with Base team warning of potential slip to 2027 if FOCIL is added alongside ePBS. The chapter's structural argument is that MEV-Boost (the out-of-protocol PBS) has been the de facto enshrined PBS for four years and counting; the in-protocol replacement is still ~12+ months out. (EF Checkpoint #9; Blockonomi — Justin Drake's Strawmap; Phemex — Glamsterdam delays; accessed 2026-05-14)

Exclusive order flow on Ethereum (Ch 7 anchor; do not re-litigate)

  1. Wu et al. identified 75 Exclusive Order Flow arrangements on Ethereum accounting for ~71% of trading-related builder revenue. Pahari and Canidio's December 2024 analysis (8-minute window, 15,097 blocks via Agnostic Relay) found that exclusive transactions account for 77.2%–84% of total fees paid in winning Ethereum blocks. Only ~7% of exclusive transaction value originated from senders routing exclusively to one builder, meaning most exclusivity is dynamic builder-routing rather than contractual. Banana Gun (paired with Titan; ~37.1% of estimated proposer losses from EOF) and Maestro (paired with Beaverbuild; ~6%) are the two largest named arrangements; Banana Gun's user fee → Titan retention → proposer payment trace was 4,466.89 ETH → 2,915.65 ETH → 2,271.26 ETH retained by Titan (Ch 7). Ch 10 references; full development in Ch 7. (Wu et al. — arXiv 2405.01329; Pahari & Canidio — arXiv 2509.16052; accessed 2026-05-14)

The L2 landscape in May 2026

  1. Total L2 TVL is approximately $48 billion across ~73 active rollups. Concentration is sharp: Arbitrum, Base, Optimism handle ~90% of L2 transaction volume. TVL rankings (May 2026):

    • Arbitrum One ~$14.9–16.9B (40–44% of L2 TVL)
    • Base ~$10.7–11.2B (~28–33%)
    • OP Mainnet ~$5.6B
    • zkSync Era ~$4.1B
    • Linea ~$3.4B
    • Scroll ~$2.1B
    • Starknet ~$1.5B
    • Unichain ~$0.88B Arbitrum + Base = ~77% of all L2 DeFi liquidity. Stage 1 (permissionless fraud proofs + trustless exit) status has now been achieved by Arbitrum One, Base, OP Mainnet, Starknet, Scroll, and Ink. No L2 has reached Stage 2 (fully decentralised governance) as of May 2026. (L2BEAT — Stages; L2BEAT TVS; Spoted Crypto — L2 Comparison 2026; accessed 2026-05-14)

  2. Daily L2 transactions in 2026 (approximate):

The L2 sequencer economy — concentrated, centralized, and revenue-bearing

  1. Every major L2 in production in May 2026 operates with a single centralized sequencer. Arbitrum One: sequencer operated by Offchain Labs. Base: sequencer operated by Coinbase. OP Mainnet: sequencer operated by OP Labs. Unichain: sequencer operated by Uniswap Labs. Linea: sequencer operated by Consensys (Maru client / QBFT upgrade replaced earlier fully centralised sequencer with multi-node consensus, but operators remain Consensys-affiliated). The single-sequencer architecture concentrates transaction ordering, MEV extraction, and fee revenue at the operator — the L2 sequencer is structurally analogous to a single block-builder with no proposer-side competition. (Eco Support — Ethereum L2 Sequencers; Unchained — What are sequencers; accessed 2026-05-14)

  2. Combined L2 sequencer revenue runs $150–250M annually in 2025–2026, with Base capturing the majority. Specifically:

    • Base: average ~$185,291/day = ~$67M/year; full-year 2025 revenue ~$870M cumulative (the L2 grew 30× year-over-year), with $51.1M in a recent month and the past 30-day on-chain fee revenue ~$2.72M. Base ≈ 62% of all L2 fee revenue, 46.6% of L2 DeFi TVL, 70% of L2 active addresses. Only L2 that was profitable in 2025 (~$55M net). Base sequencer priority fees alone average ~$156,138/day = ~86.1% of daily sequencer revenue; 64.9% of all priority fees on Base came from just 250 addresses in one year. (Yellow Research — Ethereum L2 Fee Revenue Competition 2026; CoinLaw — L2 Gas Fee Statistics; Pine Analytics — Compression of L1 Value Capture; accessed 2026-05-14)
    • Arbitrum: gross DAO income ~$26M annualized as of Q4 2025; Timeboost generated ~$3M in its first three months and now contributes ~$3M/year run-rate (about 26% of DAO income). [Blockworks — Timeboost; DL News — Timeboost; The Block — Timeboost $2M.]
    • Optimism + OP Stack: per Messari, lower per-chain revenue but with Superchain shared sequencer in progress.

  3. Arbitrum Timeboost is the canonical published L2-side EOF-equivalent surface in 2026. Launched April 2025, Timeboost is a sealed-bid second-price auction selling 200-millisecond "express lane" priority for a 60-second slot. Bidders compete to win the express lane; the auction's second-highest bid routes to the Arbitrum DAO (97% of revenue) and (3% to the auctioneer). The express-lane auction now accounts for 20–30% of daily DEX volume on Arbitrum. Empirical findings from Castro et al. (arXiv 2509.22143, September 2025): two entities win >90% of Timeboost auctions; competition has declined over time; revenue has trended downward as auction concentration tightened. The chapter's L2-side EOF anchor: the same exclusivity pattern Ch 7 documented on L1 has re-emerged at the sequencer layer on L2. (Arbitrum Docs — Timeboost; DL News — Arbitrum scooped up $3M; Castro et al. — arXiv 2509.22143; Blockworks — Timeboost goes live; accessed 2026-05-14)

  4. Shared / decentralised sequencers are in production but not yet at material scale. Espresso Systems' Espresso Sequencer launched mainnet on 12 February 2026 with HotShot consensus and sub-second finality; first integrations with Caldera / Arbitrum Orbit chains; throughput target 25 MB/s (from current 5+ MB/s). Astria's shared sequencer is in production for modular EVM rollups. Optimism's Superchain shared sequencer (in collaboration with Espresso + OP Labs) is targeted for 2026 production alongside native interop. Conduit (the largest RaaS provider, ~20% of mainnet rollups) and Caldera support both Arbitrum Orbit and OP Stack with built-in Espresso/Astria integration. As of May 2026 no major standalone L2 has decentralised its sequencer in production at scale; the realistic horizon is late 2026 to 2027. (Espresso Systems — Caldera collaboration; Astria — Shared Sequencer Network; Optimism — Welcoming Unichain; DiaData — RaaS map / Conduit; accessed 2026-05-14)

The L2 → L1 settlement layer and the blob market

  1. EIP-4844 (Proto-Danksharding, Dencun, 13 March 2024) introduced the blob data lane for rollups. Pectra (May 2025) doubled blob capacity (target 3→6, max 6→9). Fusaka (3 December 2025) introduced PeerDAS (EIP-7594) and EIP-7918, reducing per-node blob bandwidth ~8×. Two follow-up "Blob Parameter Only" (BPO) forks pushed the per-block blob target from 3 to 14 and the maximum from 6 to 21 over Dec 2025–Jan 2026. Long-term goal under full Danksharding is ~128 blobs per block. The fee mechanics adjust ~8.2% up under congestion / ~14.5% down under scarcity. Industry estimates suggest blob fees could contribute 30–50% of total ETH burn by 2026 depending on L2 activity. L2-to-L1 blob fee payment collapsed from ~$113M (2024) to ~$10M (2025) — a >90% decline — while L2 fee revenue grew. The L2 architecture compressed L2's payment to L1 by an order of magnitude. (EIP-4844 spec; Zeeve — Fusaka explainer; DataWallet — EIP-4844 Explained; Pine Analytics — L1 Value Capture; accessed 2026-05-14)

Cross-chain MEV — the new extraction surface

  1. The Cross-Chain Arbitrage paper (Maire, Sviridov, Capponi, Wattenhofer — Flashbots + TU Munich; ACM SIGMETRICS 2024 / arXiv 2501.17335) is the chapter's quantitative anchor. From September 2023 through August 2024, the researchers identified 242,535 cross-chain arbitrages across nine blockchains, totaling ~$868.64M of trading volume and generating ~$10.05M searcher revenue / $8.65M net profit. Activity grew 5.5× over the study period and surged after Dencun (March 2024) cut rollup costs. Composition: 58.35% L1↔L2 pairs (driven by native bridge availability), 35.67% L2↔L2 (relying on third-party bridges or pre-positioned inventory). Most trades (66.96%) use pre-positioned inventory and settle in ~9 seconds; bridge-based arbitrages take ~242 seconds. Concentration: top five addresses execute >50% of all trades; one address captures ~40% of daily post-Dencun volume. This is the chapter's "L2 fragmentation creates a new extraction surface" empirical anchor. (Maire et al. — arXiv 2501.17335; Flashbots Collective — Detecting Cross-Chain Arbitrages; accessed 2026-05-14)

  2. Cross-rollup MEV is now a measured non-atomic extraction surface, distinct from atomic L1 MEV. Gogol et al. (arXiv 2406.02172, October 2024; some authors at Matter Labs at time of research) quantified non-atomic MEV between Arbitrum, Base, Optimism, and zkSync Era. Findings: >500,000 unexplored arbitrage opportunities identified. Opportunities persist for 10–20 blocks on average. Arbitrage opportunities are 0.03%–0.05% of trading volume on Arbitrum, Base, Optimism; ~0.25% on zkSync Era. Atomicity is the structural difference: the multi-chain world is not very atomic, which creates risk surfaces searchers must price into bids. Shared sequencers (Espresso, Astria) and intent-based architectures (UniswapX cross-chain, CoW Swap) attempt to re-introduce atomicity at the sequencer or solver layer; neither has replaced bridge-based execution at scale. (Gogol et al. — arXiv 2406.02172; Blocknative — Fundamentals of Cross-Chain MEV; accessed 2026-05-14)

Restaking — EigenLayer as the third extraction surface

  1. EigenLayer's restaking TVL is approximately $15.3–19B across 4.36–4.6M ETH at ~1,900 active operators in early-to-mid 2026. ~94% market share of the restaking ecosystem. All-time high TVL ~$19.7B (some sources cite ~$25B aggregate as restaking-plus-Eigen-Cloud). Slashing officially went live 17 April 2025, making EigenLayer "complete." ELIP-006 Redistributable Slashing (mainnet, 2026) introduces Redistributable Operator Sets — slashed funds either burned (standard) or redistributed (redistributable). AVS roster (named in 2026): EigenDA (data availability, the largest single AVS by restaked stake), AltLayer MACH (fast finality for OP Mainnet and Arbitrum One, bundled with Caldera/Conduit/Gelato/AltLayer RaaS), Hyperlane (cross-chain messaging), Witness Chain (DePIN proof-of-location), plus a "Vertical AVS" trend toward specialized verification (AI model evaluation across 280+ crypto-AI projects). (BlockEden — EigenLayer Crosses $18B; Mitosis University — $25B restaking economy; Eigenlayer App — AVS roster; Blockdaemon — Mainnet slashing; accessed 2026-05-14)

  2. Restaking is structurally the cross-chain extraction surface's third layer. An Ethereum-staked validator now secures arbitrary AVSs, some of which operate cross-chain (Hyperlane is the named example; AltLayer MACH provides fast finality across Arbitrum + OP Mainnet from a single Ethereum-secured operator set). The economic implication: the same restaked ETH that earns base PoS rewards now also earns AVS rewards, and is exposed to slashing across multiple operator sets simultaneously. The L2 fee economy + the restaking surface + PBS exclusive flow form the chapter's "PBS + L2 fragmentation + restaking trifecta" — the three coexisting extraction surfaces that define Ethereum's structural shape in 2026. (EigenLayer Docs — Restaking Overview; BlockEden — Vertical AVS; accessed 2026-05-14)

The Hyperliquid Arbitrum bridge as L2 thesis hint

  1. Hyperliquid's Arbitrum bridge — the chain's I/O surface in 2024–2025, the site of the December 2024 DPRK-tagged-wallet scare (~$502.71M single-day outflow) — is being deprecated in favour of natively-minted USDC via Circle's CCTP. Native USDC linked between HyperCore and HyperEVM on 8 December 2025; CCTP enables one-click deposits from any of >20 CCTP-supported chains; "in the final state, the Arbitrum bridge will be deprecated and all USDC will be natively minted." Ch 9 anchor; Ch 10 uses this as a forward-looking signal: the Arbitrum infrastructure that handled most of Hyperliquid's USDC I/O is being replaced not because Arbitrum's security is inadequate but because its structural role — bridging — has been compressed by native cross-chain protocols. The "L2 as I/O surface" use-case is itself being abstracted away. (Hyperliquid on X — bridge deprecation; CCN — Hyperliquid Retires Arbitrum Bridge; StableDash — HyperCore/HyperEVM native USDC; accessed 2026-05-14)

Censorship resistance, regulation, ETH ETF staking

  1. OFAC sanctions on Tornado Cash were lifted on 21 March 2025 following the Fifth Circuit Court of Appeals ruling (November 2024) that OFAC had exceeded its statutory authority under IEEPA — the court held that immutable smart contracts did not meet the statutory definition of "property." Treasury delisted the protocol's smart contracts. The Roman Storm trial proceeded on 14 July 2025 on money-laundering / sanctions-violation / unlicensed-money-transmitter charges (separate from the smart contract delisting). The chapter must update Ch 5's "Coinbase Cloud OFAC filtering" framing: the OFAC issue at the Ethereum builder/validator layer is now narrower than it was in 2022–2024, though Coinbase Cloud still offers OFAC SDN filtering as an optional service to enterprise clients (per Q1 2026 Coinbase Validator Performance Report). (Treasury — Tornado Cash Delisting; CoinDesk — US Government Removes Tornado Cash Sanctions; Venable LLP — Treasury Lifts Sanctions on Tornado Cash; Coinbase — Q1 2026 Ethereum Validator Performance Report; accessed 2026-05-14)

  2. The SEC + CFTC's 17 March 2026 joint interpretive release confirmed that protocol staking of non-security digital commodities (including ETH) does not trigger Securities Act registration requirements — applying to solo, custodial, and liquid staking. BlackRock's iShares Staked Ethereum Trust (ticker ETHB) launched on 12 March 2026 with ~$107M seed capital; Grayscale ETHE began offering staking rewards. Effective net staking yield to ETF investors ~1.9–2.6% after 0.50–0.95% management fees (vs ~3.1–3.3% gross). The chapter's structural finding: Ethereum's exclusive-flow surface is now the most regulatory-attentive on-chain surface in the book, with Bitwise's HYPE ETF approving (Chs 5/7/9), Coinbase Cloud's optional OFAC filtering, EU PFOF ban effective 30 June 2026 (Ch 7), and SEC withdrawal of Order Competition Rule + Regulation Best Execution (12 June 2025, Ch 7). Ethereum is the chain whose extraction architecture is most studied and regulated; whether that produces meaningfully different outcomes for retail than Solana's vertical integration or Hyperliquid's in-consensus matching is the chapter's open question. (CoinDesk — BlackRock Debuts Staked Ether ETF; Bitget — Ethereum Staking ETFs 2026; Phemex — SEC Crypto Ruling 2026; accessed 2026-05-14)

The trifecta — chapter's verdict frame

  1. Ethereum's structural shape in 2026 is the PBS + L2 fragmentation + restaking trifecta. The chapter's load-bearing claim, building on the prior 21 claims:

    • PBS (Ch 5 + Ch 7): Titan ~52%, BuilderNet ~25%, Quasar ~15% on builder; the most explicit exclusive-flow surface in the book (Wu et al., Pahari-Canidio); ePBS slipping to 2026–2027.
    • L2 fragmentation (Ch 10): Arbitrum, Base, Optimism = ~90% of L2 transaction volume; Base captures ~62% of L2 fee revenue at the Coinbase-operated sequencer; single-sequencer architectures concentrate fee + MEV revenue at the operator; cross-chain MEV is a measured ~$10M/year extraction surface.
    • Restaking (Ch 10): EigenLayer ~$15–19B / 4.6M ETH across ~1,900 operators; slashing live; AVSs include EigenDA, AltLayer MACH, Hyperlane, Witness Chain; the cross-chain extraction surface restaking unlocks is the third layer.

    The three surfaces interact: an exclusive-flow searcher on L1 may also be a Timeboost express-lane winner on Arbitrum, and a top-five cross-chain arbitrageur using pre-positioned inventory between Base ↔ Arbitrum ↔ L1, and indirectly relying on EigenDA for data availability. The same five-to-ten firms operate across all three surfaces; the architecture's explicit separation of concerns has not produced a corresponding separation of operators. This is the chapter's structural verdict.

  2. The trifecta produces a different value-capture pattern from Solana's compressed vertical integration (Ch 8) or Hyperliquid's eliminated-surface in-consensus matching (Ch 9). Solana concentrates value capture into the validator+builder+searcher stack (vertical). Hyperliquid concentrates it into the HLP + validator set + HIP-3 deployers + HYPE buyback machine (horizontal but small-actor-count). Ethereum disperses value capture across L1 PBS + L2 sequencers + L2-to-L1 blob payments + restaking-secured AVSs — but the dispersion is along architectural lines (PBS, L2, restaking), not along operator lines. Lido (24% of stake), Coinbase (5–12% depending on metric), and Kiln (6%) are the named L1 stake operators; Coinbase Cloud + Offchain Labs + OP Labs are the named L2 sequencer operators; Titan + BuilderNet + Quasar are the named builders; EigenDA + AltLayer + Hyperlane are the named AVSs. The same firms — Coinbase, Consensys, Flashbots, and a handful of others — appear at multiple layers. The architecture is more legible than Solana's or Hyperliquid's; the operator concentration is comparable.

2. Numbers to verify ​

#NumberSourceDateFlag
N1Ethereum validator count May 2026: ~1.06–1.1M; total staked ETH ~35.86M (~28.91% of supply)CoinLaw; Datawallet; beaconcha.inMay 2026Strong; live numbers will drift on draft date
N2Pectra (7 May 2025 Epoch 364032) EIP-7251: max effective balance 32 → 2,048 ETH; ~40% network message overhead reductionCoinDesk; Consensys7 May 2025Already cited Ch 5
N3Ethereum L1 daily gas revenue: peak ~$23M/day → ~$6.3M/day in 2025; mildly inflationary ~0.23%/year post-Pectra; daily ETH burn ~3.26 ETH/day post-Pectra (−71%)CoinLaw; Bitcoin Ethereum News2025Strong; the "ultrasound money" contestation
N4Hyperliquid Q1 2026 ~$144.8M (#1 crypto app); Solana Q1 2026 chain REV ~$89.5M (#2); Ethereum Q1 2026 chain fees ~$82MCoinpedia on X; PANews Solana Q1 Report; Odaily — L1 Value Capture ShrinksQ1 2026Strong; structural anchor
N5MEV-Boost builder shares early 2026 (relayscan.io 24h): Titan ~52.16% / BuilderNet ~24.63% / Quasar ~15.06% / Eureka ~2.41% / Beaverbuild ~1.81% / Bombora ~1.31%relayscan.io; bytenoob.io — BuilderNet deep diveEarly 2026Live; refresh on draft date
N6MEV-Boost relay shares May 2026: Ultrasound ~35.71%, Titan Relay ~26.09%, BloXroute Max ~13.38%, BloXroute Regulated ~12.30%, Aestus ~7.32%, Flashbots ~2.09%relayscan.ioMay 2026Live; already cited Ch 5
N7Eureka Labs: $6.7M seed (April–June 2025); Spark Capital + Collider Ventures; ~1.5% builder share; founded Dec 2024The Block; Phemex News2025New for Ch 10
N8ePBS / Glamsterdam slipping past Q2 2026; "Q2 seems to me to be unlikely" (EF Checkpoint #9, April 2026); realistic Q3/Q4 2026 with 2027 risk if FOCIL addedEF Checkpoint #9; PhemexApril 2026Already cited Ch 5; consistent
N9Justin Drake "Strawmap" (Feb 2026): seven-fork roadmap through 2029; one consensus + one execution headliner per forkBlockonomi — Strawmap; TapbitFeb 2026Strong; refers to Hegotá as next fork after Glamsterdam
N10Wu et al. — 75 EOF arrangements / ~71% of trading-related Ethereum builder revenue; Banana Gun ~37.1% of estimated proposer losses; Maestro ~6%Wu et al. arXiv 2405.013292024–2025Already cited Ch 7
N11Pahari–Canidio — 77.2%–84% of fees in winning Ethereum blocks come from exclusive transactions; ~7% from senders routing exclusively to one builderPahari & Canidio arXiv 2509.16052November 2025Already cited Ch 7
N12Total L2 TVL ~$48B across ~73 active rollups; top 3 (Arb/Base/OP) = ~90% of transaction volumeL2BEAT; Spoted Crypto L2 2026May 2026Strong
N13L2 TVL rankings May 2026: Arbitrum ~$14.9–16.9B / Base ~$10.7–11.2B / OP ~$5.6B / zkSync Era ~$4.1B / Linea ~$3.4B / Scroll ~$2.1B / Starknet ~$1.5B / Unichain ~$0.88BL2BEAT; Spoted CryptoMay 2026Strong; live
N14L2 daily transactions: Base ~12.89M, Arbitrum ~4.3M, Starknet ~585K, Linea ~211K; L2 ecosystem ~1.9M+ daily 2025; L2s now exceed Ethereum mainnet by wide margin in 2026CoinLaw L2 Adoption; Dune; Yellow2025–2026Strong
N15L2 stage classification May 2026: Stage 1 = Arbitrum One, Base, OP Mainnet, Starknet, Scroll, Ink; no L2 yet Stage 2L2BEAT StagesMay 2026Live
N16L2 sequencer revenue combined ~$150–250M/year; Base ~62% of L2 fee revenue, ~$185K/day = ~$67M/year, ~$870M lifetime, $51.1M recent month, ~$55M net profit in 2025 (only profitable L2)Yellow Research; Pine Analytics; CoinLaw L2 Gas2025–2026Strong
N17Base priority fees ~$156K/day = 86.1% of daily sequencer revenue; 64.9% of priority fees from just 250 addressesCoinLaw L2 Gas2025–2026Strong; chapter's key concentration figure
N18Coinbase Q1 2026 on-chain revenue $109.4M; Base 30-day on-chain fee revenue $2.72M; Coinbase Q1 2026 total revenue $1.4B, net loss $394MCoinbase Q1 2026 10-Q; Daily Political — Coinbase Q1 2026Q1 2026Primary
N19Arbitrum Timeboost: launched April 2025; ~$3M in fees first 3 months; ~$3M/year run-rate by March 2026 (26% of DAO income); 20–30% of daily DEX volume on Arbitrum; >90% of auctions won by 2 entities (Castro et al.)DL News; Castro et al. arXiv 2509.22143; Blockworks2025–2026Strong; chapter's L2-EOF anchor
N20Espresso Sequencer mainnet 12 Feb 2026; sub-second finality; throughput target 25 MB/s (from 5+ MB/s); Caldera + Arbitrum Orbit integrationEspresso Caldera Medium; Bitget — Espresso Price PredictionFeb 2026Strong
N21EIP-4844 (Dencun 13 March 2024); Pectra (May 2025) doubled blob capacity 3→6 target, 6→9 max; Fusaka (3 Dec 2025) PeerDAS EIP-7594 + EIP-7918 reduced node bandwidth ~8×; BPO1 (17 Dec 2025) 6→~10 target; BPO2 (7 Jan 2026) ~14 target, 21 maxEIP-4844; Zeeve Fusaka; DataWallet; DEV Community EIP-78922024–2026Strong
N22L2-to-L1 fee payment: ~$113M (2024) → ~$10M (2025); >90% decline post-Pectra/Fusaka; ETH burn from blob fees projected 30–50% of total by 2026Pine Analytics; DataWallet2024–2026Strong
N23Cross-Chain Arbitrage: 242,535 arbitrages over Sep 2023–Aug 2024 across 9 chains; ~$868.64M volume; ~$10.05M revenue / ~$8.65M profit; 58.35% L1↔L2, 35.67% L2↔L2; pre-positioned 66.96%/9s vs bridge 242s; top 5 addresses >50% of trades; 1 address ~40% post-DencunMaire et al. arXiv 2501.17335; Flashbots Collective2024–2025Chapter's load-bearing cross-chain MEV anchor
N24Cross-Rollup MEV (Gogol et al.): >500K unexplored opportunities; 0.03%–0.05% of volume on Arb/Base/OP; ~0.25% on zkSync Era; opportunities persist 10–20 blocksGogol et al. arXiv 2406.02172October 2024Strong
N25EigenLayer TVL May 2026: ~$15.3B (4.36M ETH) per Fensory, ~$18B per BlockEden March 2026, ~$19B / 4.6M ETH per other sources, ~$25B aggregate (with Eigen Cloud) per Mitosis; ~1,900 active operators; ~94% restaking market share; all-time high ~$19.7BBlockEden; Fensory; Mitosis UniversityMarch–May 2026Range; pin a single snapshot for the chapter
N26EigenLayer slashing live since 17 April 2025; ELIP-006 Redistributable Slashing mainnet; redistributable vs standard operator setsBlockdaemon; Gate.com — EigenLayer Slashing LiveApril 2025Primary
N27AVS landscape: EigenDA (DA, largest AVS), AltLayer MACH (OP Mainnet + Arbitrum fast finality), Hyperlane (cross-chain messaging), Witness Chain (DePIN), AI-verification AVSs (280+ projects)Eigenlayer App; DAIC Capital; Onchain Times2026Strong
N28Lido stake share ~23–24% (8.72M ETH); Coinbase ~5.1% (1.84M ETH) by network share / 12.17% by cumulative validator share; Kiln 6% Ethereum; total top-10 entities ~60% concentrationLido Scorecard; Datawallet; news.bitcoin.com — Coinbase 4.5M ETH2026Already cited Ch 5; consistent
N29Hyperliquid Arbitrum bridge deprecation: native USDC linked HyperCore↔HyperEVM 8 Dec 2025; CCTP 20+ supported chains; "in the final state, the Arbitrum bridge will be deprecated"; record Dec 2024 single-day outflow ~$502.71M during DPRK-tagged wallet scareHyperliquid on X; CCN; StableDash 8 Dec 2025; CointelegraphDec 2024 – Dec 2025Already cited Ch 9; consistent
N30OFAC sanctions on Tornado Cash lifted 21 March 2025 following Fifth Circuit Nov 2024 ruling; Treasury delisting; Roman Storm trial 14 July 2025 on separate chargesTreasury — Tornado Cash Delisting; CoinDesk; Venable LLPMarch 2025Strong
N31BlackRock ETHB launched 12 March 2026 with $107M seed; SEC+CFTC 17 March 2026 joint interpretive release on protocol staking; ETF management fees 0.50–0.95%; effective net yield ~1.9–2.6% vs gross ~3.1–3.3%CoinDesk — BlackRock ETHB; Bitget — Ethereum Staking ETFs 2026; Phemex — SEC Crypto RulingMarch 2026Strong
N32Coinbase Cloud Q1 2026: 4.5M ETH staked (~12.17% of network or 5.1% by other measure); 99.98% participation; zero slashing since inception; 7 MEV relays connected (Flashbots, BloXroute Max + Regulated, Ultrasound, Agnostic, Aestus, Titan); optional OFAC SDN filteringCoinbase ETH Validator Q1 2026 Report; news.bitcoin.com — 99.98% uptimeQ1 2026Already cited Ch 5; consistent

Deliberately not pinned:

  • Aggregate cross-chain MEV in 2025–2026 (the Maire et al. paper covers Sep 2023–Aug 2024; 2025–2026 may be materially larger but no comparable academic dataset published yet)
  • EigenLayer AVS-revenue-to-operator dollar share in 2026 (small relative to TVL; no clean aggregated figure)
  • Total Quasar revenue / margin / corporate ownership (the firm is comparatively opaque; founders not publicly named; no published margin)
  • Per-rollup Maker-Taker fee breakdown for Base / Arbitrum / Optimism MEV vs sequencer fee components beyond the Base "priority fees ~86.1%" anchor
  • The cumulative dollar Eureka Labs has captured under "programmable blocks" (firm is recent + small share)
  • Specific operator P&L on AltLayer MACH / Hyperlane / Witness Chain (not in primary sources)

3. Contested or evolving claims ​

  • EigenLayer TVL range. Sources cite values from $15.258B (Fensory, early 2026) to $18B (BlockEden, March 2026) to $19B (current sources) to $25B (Mitosis aggregation including Eigen Cloud). The chapter should pin a single dated snapshot (e.g., "~$18B as of March 2026 per BlockEden / Vertical AVS report" or "~$15–19B range, depending on whether the figure includes EIGEN-token-secured stake and Eigen Cloud's separate TVL"). The structural argument (EigenLayer is the dominant restaking primitive; ~1,900 operators; ~94% market share) is durable; the dollar figure is moving and depends on accounting.

  • Hyperliquid's Solana ranking durability. Ch 9 anchors the Q1 2026 reversal (Hyperliquid #1, Solana #2, ~$144.8M vs ~$89.5M). Whether Solana re-passes Hyperliquid in Q2 2026 (Alpenglow testnet 11 May 2026; potential memecoin recovery) is an open question. The chapter must frame as Q1 2026 fact, not durable shape — consistent with Ch 9's framing.

  • The "ultrasound money" framing for ETH. Pre-Dencun: ETH was deflationary; post-Dencun: ETH became slightly inflationary. The chapter must frame this as a deliberate trade-off (the L2 architecture moved fee economy off L1; the resulting L1 inflation is a feature of the scaling design, not a bug) rather than as a structural failure. The chapter should be clinical about the trade-off — readers will have heard "ultrasound money" and need it contextualised.

  • L2 sequencer decentralisation timeline. Espresso went mainnet Feb 2026, but no major standalone L2 has decentralised its sequencer in production at material scale. The chapter should frame as "the decentralisation roadmap is real and operational, but no L2 has executed the migration at scale; the realistic horizon is late 2026 to 2027." This is consistent with Ch 12's forward-looking treatment of appchains and shared sequencing.

  • Glamsterdam timeline. EF Checkpoint #9 (April 2026) says Q2 2026 is "unlikely." Justin Drake's Strawmap (Feb 2026) outlines a 2026–2029 sequence with Glamsterdam as the next fork. Realistic mainnet activation: Q3/Q4 2026; risk of slip to 2027 if FOCIL adds complexity. The chapter must frame as "ePBS exists as a designed protocol-level fix; its mainnet activation is at least 12 months out; in the meantime MEV-Boost remains the de facto enshrined PBS."

  • The Base profitability claim. "Only L2 profitable in 2025 (~$55M net)" is a structural finding but depends on which costs are attributed. Coinbase's L2-specific cost allocation isn't fully disclosed; some L2 costs are absorbed by Coinbase's broader infrastructure. The chapter should frame as "Base is the only L2 disclosing net profitability in 2025" rather than as the definitive empirical claim.

  • Coinbase Cloud OFAC filtering. Brian Armstrong's prior statement was that Coinbase would "rather cease staking than censor"; Coinbase's actual product offering includes optional OFAC filtering for enterprise clients. The framing in Coinbase's Q1 2026 Validator Performance Report: "OFAC screening is available as an option for customers who need transaction filtering." The chapter should frame as: "Coinbase Cloud offers OFAC filtering as an optional service to a subset of enterprise clients; the company's published position is that it would not adopt filtering as a default." Note the framing in the post-Tornado-Cash-delisting (March 2025) regulatory environment is narrower than pre-March-2025.

  • The Quasar founders / corporate structure. Quasar holds ~15% of Ethereum block construction but founders are not publicly named in any source I could pin. The chapter can note its share but should frame the firm's ownership as opaque, distinct from Titan (Kubi Mensah) and BuilderNet (Flashbots + Beaverbuild + Nethermind).

  • Cross-chain MEV scale in 2025–2026. The Maire et al. paper covers Sep 2023–Aug 2024 with $868.64M cumulative volume / $10.05M revenue. Activity grew 5.5× over that period. Extrapolating to 2025–2026 implies materially larger volumes but no academic dataset has yet been published at that scale. The chapter must frame as "the most-cited published cross-chain MEV study covers 2023–2024 and shows a 5.5× growth trajectory; 2025–2026 data is not yet published at comparable rigor but operator behaviour (Wintermute, Jump, GSR named as cross-chain searchers in industry commentary) suggests the surface has continued to grow."

  • The Wintermute "compression" claim from Ch 9. May 2026 reports indicate Wintermute's BTC + ETH resting liquidity on Hyperliquid dropped from ~$40M to ~$4M. Ch 9 flagged this. For Ch 10's purposes the relevant point is that Wintermute's cross-chain MM operation includes Ethereum L2s and Hyperliquid; the operator concentration story is durable even when individual venue allocations move.

4. Characters introduced ​

No new "Meet the Actor" sidebar. The chapter is structurally a chain-comparison chapter; the major actor types (validator, builder, sequencer, restaker, searcher) are all introduced in Chs 4–7. Ch 10's job is to develop the integration — how the same firms operate across PBS + L2 sequencer + restaking + cross-chain MEV — not to introduce new actor categories.

Returning institutions / actors, with Ch 10-specific roles:

L1 PBS layer (Ch 5 + Ch 7 anchors):

  • Titan Builder — ~52% builder share; Kubi Mensah-led; Banana Gun exclusive (Ch 7); the canonical EOF case.
  • BuilderNet — ~25% share; Flashbots + Beaverbuild + Nethermind operators; the explicit non-exclusive alternative.
  • Quasar — ~15% share; founders not publicly named; "neutral, non-censoring builder" per own marketing.
  • Eureka Labs — ~1.5% share; "programmable blocks" surface; Spark Capital + Collider Ventures backed; new for Ch 10.
  • Coinbase Cloud — 4.5M ETH staked, ~12.17% network share; the worked-example anchor from Ch 5; OFAC-optional; reaches Ch 10 as L1 stake holder + Base sequencer operator (the dual-layer anchor).
  • Lido + Curated Operator Set — 24% of stake; ~36 named operators including Chorus One, P2P, Stakefish, Kiln, Allnodes, Figment (Ch 7).
  • Kiln — 6% of Ethereum, 2.5% of Solana; institutional staking-as-a-service across 50+ networks (Ch 5).
  • Flashbots — operator of BuilderNet; SUAVE archived; MEV-Blocker transferred to Consensys SMG Jan 2026 (Ch 6); maintained Ethereum L1 PBS infrastructure for 4+ years.

L2 sequencer layer (new for Ch 10):

  • Offchain Labs — Arbitrum sequencer operator; Timeboost (the L2 EOF-equivalent surface).
  • OP Labs — OP Mainnet sequencer operator; Superchain shared-sequencer collaboration with Espresso.
  • Coinbase (via Base) — Base sequencer operator; ~62% of L2 fee revenue; ~$185K/day sequencer revenue.
  • Uniswap Labs — Unichain sequencer operator; Flashbots collaboration on TEE-based block builder targeting 250ms confirmations.
  • Consensys (via Linea) — Linea sequencer operator; Maru / QBFT upgrade; dual-burn mechanism (20% sequencer fees burn ETH, 80% buy back and burn LINEA).
  • Espresso Systems — shared sequencer operator; mainnet 12 Feb 2026; HotShot consensus; sub-second finality; Caldera + Arbitrum Orbit integrations.
  • Astria — modular EVM-agnostic shared sequencer; in production for select rollups.
  • Conduit — RaaS provider; ~20% of mainnet rollups; supports Orbit + OP Stack with built-in Espresso/Astria integration.
  • Caldera — RaaS provider; Espresso testnet collaborator; Metalayer interop layer.

Restaking layer (new for Ch 10):

  • EigenLayer / Eigen Cloud — ~$15–19B restaking TVL; ~1,900 operators; ~94% market share; slashing live April 2025.
  • EigenDA — the largest single AVS by restaked stake; data availability primitive; secures multiple OP Stack and Arbitrum Orbit chains.
  • AltLayer MACH — fast-finality AVS for OP Mainnet + Arbitrum One; bundled with Caldera / Conduit / Gelato RaaS.
  • Hyperlane — cross-chain messaging AVS; the named cross-chain extraction example.
  • Witness Chain — DePIN proof-of-location AVS.
  • Lido V3 / stVaults — restaking-adjacent; Kiln as approved node operator; institutional layer for ETH staking-plus-restaking yield.

Cross-chain MEV / extraction layer:

  • The five-address concentration — Maire et al. shows top-5 addresses execute >50% of cross-chain arbitrages, one captures ~40% of post-Dencun volume. The chapter should note the concentration as analogous to L1 builder concentration.
  • Wintermute / Jump / GSR / Tower — cross-chain searchers per industry commentary; named at L1 (Ch 4–7) and operating across L2s + Hyperliquid; specific cross-chain allocations not disclosed.
  • Banana Gun + Titan / Maestro + Beaverbuild — Ch 7 anchors; carries through to Ch 10 as L1 EOF.

Regulatory anchors:

  • OFAC / Treasury — Tornado Cash delisted 21 March 2025; the OFAC story is materially different from 2022–2024.
  • Fifth Circuit Court of Appeals — November 2024 ruling that prompted the Tornado Cash delisting.
  • SEC + CFTC joint interpretive release — 17 March 2026 confirming protocol staking does not trigger Securities Act registration.
  • BlackRock + ETHB — staked Ethereum ETF launched 12 March 2026.
  • Grayscale + ETHE — added staking rewards 2026.
  • REX-Osprey + ESK — spot Ethereum with staking ETF.

Hyperliquid bridge (Ch 9 anchor, Ch 10 forward-looking signal):

  • Circle / CCTP — the cross-chain transfer protocol replacing the Arbitrum bridge; 20+ supported chains; the structural abstraction layer that is replacing L2-based I/O for cross-chain liquidity.

5. Worked example candidates ​

The chapter is structurally a comparison/synthesis chapter. The worked example should illustrate the trifecta (PBS + L2 fragmentation + restaking) in a way that integrates rather than dissects.

Candidate A — Alice's $10,000 USDC → ETH swap, routed through the Ethereum 2026 stack ​

The Alice convention from prior chapters runs the same baseline trade on Ethereum. Phantom-equivalent (MetaMask, Rabby) wallet → Flashbots Protect / MEV-Blocker private RPC → UniswapX / CoW Swap solver auction → Beaverbuild-via-BuilderNet → Coinbase Cloud (or Lido-via-Kiln) proposer signs the block. Comparison with Ch 8's Solana ($8–15) and Ch 9's Hyperliquid ($5.50) per-trade trace demonstrates the dispersion of the Ethereum stack: more layers, lower per-layer take, but cumulative comparable to Solana 2026.

LayerActorTakeCumulative
WalletMetaMask / Rabby$0–2 (varies by wallet model)$0–2
Private RPCFlashbots Protect / MEV-BlockerRefund-positive; user often receives back portion of MEV$0 net
Solver auctionUniswapX (SCP / Wintermute)~5–15 bps depending on size~$5–15
Build + bidTitan / BuilderNet / QuasarBuilder margin (~17.75% on EOF transactions; lower on competitive flow)~$5–15
ProposeCoinbase Cloud or Lido operatorMEV-Boost bid net of relay fee(in builder bid)
SettleEthereum L1 finality (~12.8 min full economic)Gas + priority fee~$2–10 (varies with congestion)
Total~$10–30 on Ethereum L1 in 2026

Plus L2 alternative: same trade on Base would route Phantom → MetaMask → Aerodrome / Uniswap V4 → Coinbase sequencer (which captures the entire stack take at one operator), total ~$0.50–2 (the cost compression from L2s) — but with Coinbase as the single point of MEV extraction and sequencer revenue capture.

  • Pros: Mirrors Ch 8 / Ch 9 structure; illustrates the trifecta in action; shows the dispersion-vs-concentration trade-off vs Solana / Hyperliquid; integrates Titan + Coinbase + L2 sequencer.
  • Cons: Less narratively vivid than Ch 7's Banana Gun + Titan trace; risks repeating Ch 7's worked example.

Candidate B — A cross-chain searcher executing the Maire et al. pattern ​

Anchor the chapter on a specific cross-chain searcher walking through one of the 242,535 arbitrages: a price gap opens between Arbitrum and Base on a major asset (ETH, USDC, or a longer-tail token). The searcher with pre-positioned inventory closes the gap in 9 seconds; a competitor relying on a third-party bridge takes 242 seconds and loses the opportunity. The chapter walks the economics: the searcher's pre-positioned capital + the bridge-based cost structure + the L2 sequencer-controlled MEV margin + the L1 settlement cost (paid in blob fees) + the EigenLayer-secured Hyperlane message that confirmed the cross-chain price discovery.

  • Pros: Most differentiating worked-example for the chapter; integrates all three of the trifecta surfaces; produces a vivid empirical scene (top-5 concentration, 9-vs-242-second latency).
  • Cons: Composite scenario; requires the chapter to construct it from published data rather than from a single named case.

Candidate C — Coinbase as a dual-layer actor ​

The same $10,000 trade walks through Coinbase Cloud at the L1 layer (validator / proposer) and Coinbase at the Base sequencer layer (L2 operator). The chapter follows Coinbase's revenue from the trade at both layers: as L1 validator, it earns the MEV-Boost bid + base reward; as L2 sequencer, it earns the priority fee + sequencer MEV. The same firm captures revenue at two different architectural layers, with no formal disclosure of the inter-layer accounting. The chapter's "operator concentration despite architectural separation" verdict.

  • Pros: Illustrates the chapter's structural argument (the architecture separates concerns, the operators don't); ties to the regulatory dimension (Coinbase as the named institutional anchor in both PoS validator landscape and L2 sequencer landscape); develops the public-company disclosure question.
  • Cons: Narrower than A or B; doesn't develop cross-chain MEV.

Recommendation ​

Candidate A as the worked-example anchor, mirroring Ch 8 / Ch 9 structure with Alice's $10,000 trade. Candidate B's cross-chain searcher pattern threaded as a subsection inside the mechanics (not a competing worked example but a vivid illustration of the cross-chain MEV measurement). Candidate C's Coinbase dual-layer observation woven into the verdict section as the chapter's "same operators across layers" closing observation. Three-candidate synthesis rather than choose-one.

6. Open questions for Nick ​

Q1 — Word budget. Comparable to Chs 7–9 (~5,500–7,000 words inc. footnotes). The chapter has substantial referenced material (PBS already developed in Chs 5/7; sequencer + restaking + cross-chain MEV new). Recommend 5,500–6,500 words. Acceptable, or tighter?

Q2 — Re-litigation of Ch 5/7 material. Wu et al. + Pahari-Canidio + Banana Gun + Titan share + Coinbase Cloud are all developed in prior chapters. Recommend Ch 10 references in passing and adds two new beats: (a) Eureka Labs as new builder entrant; (b) the L2-sequencer / cross-chain MEV / restaking layer as the new material. Confirm balance.

Q3 — Cross-chain MEV as primary new material. The Maire et al. (Flashbots + TU Munich) Sep 2023–Aug 2024 dataset is the strongest published cross-chain MEV academic anchor. The Gogol et al. cross-rollup paper is the L2↔L2 anchor. Together they support a substantial subsection on "cross-chain MEV as the surface created by L2 fragmentation." Confirm this is the chapter's new structural surface.

Q4 — Restaking treatment depth. EigenLayer is the dominant primitive; AVSs are real but small-revenue. Recommend ~500-word subsection covering EigenLayer + EigenDA + AltLayer MACH + Hyperlane + slashing-live status. Or do you want restaking as a forward-link to Ch 12 (forward-looking) instead? The chapter's structural argument benefits from including it (the "trifecta" frame); but it could be deferred. My preference: include here, lighter treatment in Ch 12.

Q5 — Glamsterdam / ePBS framing. Ch 5 already established that ePBS has slipped past Glamsterdam's H1 2026 target. Ch 10 reinforces with the April 2026 Checkpoint #9 admission and Justin Drake's Strawmap. Recommend a ~200-word forward-link to ePBS as "the in-protocol replacement for MEV-Boost is still 12–18+ months out, possibly into 2027 if FOCIL is added; MEV-Boost remains the de facto enshrined PBS for the chapter's purposes." Confirm framing.

Q6 — L2 sequencer-side EOF parallel to Ch 7 Ethereum-side EOF. Arbitrum Timeboost is the canonical published L2-side EOF-equivalent surface: a sealed-bid second-price auction that 90%+ of which is captured by two entities (Castro et al.). Recommend developing as a substantial subsection illustrating the L2 sequencer-as-EOF parallel — the same exclusivity pattern emerges architecturally one level down from L1 PBS. Confirm.

Q7 — Worked example. Candidate A (Alice $10K Ethereum L1 + Base L2 trace, mirroring Ch 8 / Ch 9 structure) is my recommendation. Cross-chain searcher (B) threaded as subsection. Coinbase dual-layer (C) in verdict. Confirm or override.

Q8 — The Hyperliquid Arbitrum bridge as L2 thesis hint. Ch 9 anchored the deprecation; Ch 10's role is to develop what it tells us about where the L2 thesis is going. Recommend ~200-word closing paragraph: "the Arbitrum bridge that was the chain's load-bearing I/O surface is being replaced not because Arbitrum is inadequate but because L2-based I/O is being abstracted by native cross-chain protocols (CCTP, intent-based architectures, shared sequencers). The L2 use-case is compressing in scope." Confirm framing.

Q9 — The OFAC / Tornado Cash regulatory update. March 2025 delisting materially changed the OFAC story at the Ethereum builder layer (Wahrstätter's "72% of MEV-Boost blocks are 'censored'" figure was a 2022–2024 measurement). Recommend a one-paragraph update: the regulatory environment is narrower than during 2022–2024, but Coinbase Cloud and Aestus-style "non-OFAC-filtering" relay distinctions remain. Confirm.

Q10 — The "trifecta" verdict frame. PBS + L2 fragmentation + restaking as Ethereum's structural shape in 2026 — distinct from Solana's vertical integration (Ch 8) and Hyperliquid's elimination (Ch 9). My recommended verdict: architectural dispersion, operator concentration; the same five-to-ten firms operate across all three surfaces. Confirm this is the chapter's load-bearing closing argument.

Q11 — Ethereum L1 REV / "ultrasound money" framing. The "ultrasound money" narrative is contested post-Dencun. Recommend the chapter address this directly but clinically (not as failure, but as deliberate scaling trade-off). Confirm tone.

Q12 — Linea / Consensys treatment. Linea has a published dual-burn mechanism (20% sequencer fees burn ETH, 80% LINEA buyback-and-burn) and is moving to Type 1 zkEVM in Q1 2026. Recommend ~100 words as a "the L2 dual-burn experiment" cameo, distinct from Base's no-token-yet operator-captured-revenue model. Or skip and let the L2 section focus on Arbitrum, Base, OP?

Sources cited ​

All URLs accessed 2026-05-14 unless otherwise noted. Consolidated bibliography for Phase 2.

Primary research and protocol documentation:

Primary academic research:

Primary regulatory and institutional:

News and analytical coverage:

Pulled through from prior chapters' RESEARCH.md (URLs in those files):

  • Ch 5: Pectra (May 2025); Titan + Banana Gun trace; relayscan.io 24h snapshot; ePBS slipping past Glamsterdam; Coinbase Cloud 4.5M ETH; BuilderNet launch
  • Ch 6: Helius wallet rebate model; full infrastructure stack treatment; bloXroute + Ultrasound + Aestus relays
  • Ch 7: Wu et al. 75 EOF; Pahari–Canidio 77.2%–84%; Banana Gun/Titan; Maestro/Beaverbuild; SEC withdrawal of Order Competition Rule (12 June 2025); EU PFOF ban (30 June 2026)
  • Ch 8: Solana Q1 2026 REV $89.5M; Ethereum Q1 2026 #3 ranking
  • Ch 9: Bitwise BHYP ETF launch (15 May 2026); Hyperliquid Arbitrum bridge deprecation; native USDC via CCTP; Wintermute mid-May 2026 compression flag

Phase 1 is complete. Per the user's compressed-review pattern, Phase 2 (OUTLINE.md) follows immediately after Nick's review of this RESEARCH.md.