Shared sequencing is not atomic composability. It coordinates transaction ordering across multiple rollups, but it does not guarantee atomic execution or settlement across them. This creates a critical gap where cross-rollup MEV opportunities persist.
mev-the-hidden-tax-of-crypto
Blog
Why Shared Sequencing Won't Solve Cross-Rollup MEV
Shared sequencers are a partial solution for intra-rollup ordering. The lucrative, complex MEV exists in the gaps between sovereign rollup stacks and L1s, requiring a fundamentally different architecture.
introduction
THE FLAWED PREMISE
Introduction
Shared sequencing is a necessary but insufficient solution for the systemic MEV problem spanning rollup ecosystems.
The MEV problem moves, not disappears. A sequencer like Espresso or Astria can prevent frontrunning within its domain, but cross-domain arbitrage between Ethereum, Arbitrum, and Optimism remains a fertile ground for searchers using tools like Flashbots.
Intent-based architectures solve a different problem. Protocols like UniswapX and Across abstract execution, but they rely on off-chain solvers who compete on price, internalizing MEV rather than eliminating it from the system.
Evidence: The 2023 Ethereum rollup landscape processed over $10B in cross-chain volume, creating predictable arbitrage delays that MEV bots exploit, a dynamic a shared sequencer alone cannot resolve.
thesis-statement
THE BOUNDARY PROBLEM
The Core Argument
Shared sequencing optimizes execution within a rollup set but fails at the critical cross-rollup boundary where the highest-value MEV is extracted.
Sequencing is not settlement. A shared sequencer like Espresso or Astria only orders transactions for its connected rollups. Final settlement and asset transfer between sovereign rollup ecosystems still require a separate bridge like Across or LayerZero, creating a hard coordination boundary.
MEV migrates to the weakest link. Proposers on Ethereum (or any settlement layer) capture the arbitrage opportunity between the final, settled states of different rollups. This cross-domain MEV is where the largest value, like DEX arbitrage between Uniswap on Arbitrum and Optimism, is extracted.
Shared sequencing creates a new MEV market. While it reduces internal rollup MEV, it centralizes the right to sequence cross-rollup bundles. This creates a meta-MEV auction for the shared sequencer role itself, as seen in designs like SUAVE, rather than eliminating the extractive market.
Evidence: The 90%+ of bridge volume on major L2s flows through third-party bridges (Across, Stargate), not native bridges. The sequencer does not control this finality frontier, which is where atomic cross-rollup arbitrage is settled.
key-trends
WHY SHARED SEQUENCING ISN'T ENOUGH
The Unaddressed MEV Landscape
Shared sequencers like Espresso and Astria standardize ordering within a rollup ecosystem, but fail to capture the most valuable cross-domain MEV.
01
The Cross-Domain Liquidity Problem
Shared sequencing optimizes for a single rollup family (e.g., all OP Stack chains). The real arbitrage and liquidation value exists between ecosystems (e.g., Arbitrum → Polygon → Base).
- $10B+ TVL is fragmented across incompatible rollup stacks.
- Latency for cross-rollup arbitrage remains ~12-45 seconds, dominated by L1 finality.
- This gap is where generalized intent solvers (UniswapX, CowSwap) and bridges (Across, LayerZero) currently operate.
12-45s
Arb Latency
$10B+
Fragmented TVL
02
The Sovereignty Trap
Rollups adopt shared sequencers for liveness and anti-censorship, but ceding ordering to a third-party network creates a new centralization vector and does not guarantee optimal cross-domain execution.
- Creates a meta-sequencer MEV market—value accrues to the sequencing network, not the rollup or its users.
- Does not solve for intent-based routing across heterogeneous proving systems (ZK vs. Optimistic).
- Rollups remain sequencer-level fragmented, unable to coordinate with chains on Celestia or EigenDA.
1
New Central Point
0
Cross-Stack Guarantees
03
The Verifiable Delay Function (VDF) Blind Spot
Proposals for fair ordering via VDFs (e.g., in Espresso's HotShot) only work within a single, synchronous consensus set. They cannot impose ordering across asynchronous rollup domains with different finality times.
- A VDF on Shared Sequencer A is meaningless for a transaction's position in Shared Sequencer B's queue.
- Cross-domain MEV extraction requires supranational coordination of block space, which no current shared sequencer architecture provides.
- This is the architectural opening for protocols like SUAVE, which aim to be a decentralized block builder for all chains.
Async
Domain Finality
Supranational
Coordination Needed
04
The Economic Reality: Extractable Value Migrates
Solving MEV in one layer simply pushes it to the next most flexible coordination layer. Shared sequencers internalize MEV within an L2 ecosystem, but amplify the value of cross-ecosystem coordination.
- This creates a bifurcated market: In-Ecosystem Ordering vs. Cross-Ecosystem Execution.
- The latter is larger and requires a universal mempool and commit-reveal schemes, not just a shared sequencer.
- Projects like Anoma and Flashbots SUAVE are targeting this higher-value layer directly.
Bifurcated
Market Structure
Universal
Mempool Required
SEQUENCER ARCHITECTURE
MEV Arena: Local vs. Global
Comparing the MEV capture and security models of isolated rollup sequencers versus shared sequencing networks like Espresso, Astria, and Radius.
| MEV Dimension | Local Sequencing (Status Quo) | Shared Sequencing (e.g., Espresso, Astria) | Enshrined Sequencing (Future) |
|---|---|---|---|
Cross-Rollup Bundle Execution | |||
MEV Revenue Capture Scope | Single Rollup | Network of Rollups | Entire L1 Ecosystem |
Proposer-Builder Separation (PBS) | Rollup-Dependent | Native to Sequencer Set | L1-Enforced |
Time-to-Mine (TTM) for Cross-Domain Arb |
| < 2 seconds (shared mempool) | < 1 second (atomic L1 inclusion) |
Trust Assumption for Censorship Resistance | Single Sequencer Operator | Sequencer Set (e.g., 100+ nodes) | L1 Validator Set |
Interoperability Protocol Required | Bridges (LayerZero, Axelar) | Native Atomic Composability | Native Atomic Composability |
Maximum Extractable Value (MEV) Leakage | High (to L1 searchers) | Reduced (captured in-network) | Minimal (redistributed) |
Implementation Complexity & Adoption Hurdle | Low (current standard) | High (requires rollup integration) | Extreme (requires L1 consensus change) |
deep-dive
THE DOMAIN PROBLEM
Architectural Blind Spot
Shared sequencing fails to address the fundamental MEV that arises from atomic execution across separate rollup domains.
Shared sequencing is domain-bound. It optimizes ordering within a single rollup set, like Espresso for the Arbitrum ecosystem. This creates a local optimum but ignores the larger cross-domain MEV game where value leaks between chains like Arbitrum and Optimism.
Cross-domain MEV is a coordination problem. A shared sequencer for Rollup A cannot force Rollup B's sequencer to include a dependent transaction. This creates predictable arbitrage delays exploited by searchers, a flaw that protocols like Across and Socket actively monetize.
The solution requires enforceable atomicity. True cross-rollup MEV capture needs a system that can guarantee execution across sovereign state machines. This is the architectural goal of shared sequencing successors like intent-based architectures (UniswapX, CowSwap) and omnichain protocols (LayerZero, Chainlink CCIP).
Evidence: Over $3B in bridge volume monthly flows through systems like Stargate and Synapse, creating a massive, fragmented MEV opportunity that no single sequencer network currently captures.
counter-argument
THE CROSS-CHAIN REALITY
The Rebuttal: "But Interoperability!"
Shared sequencing fails to address the core economic and security challenges of cross-rollup MEV.
Shared sequencing is intra-domain. A shared sequencer network like Espresso or Astria coordinates blocks for a set of rollups on a single L1. It does not coordinate execution across Ethereum, Arbitrum, and Optimism. Cross-rollup MEV requires atomic execution across sovereign settlement layers, which a single sequencer set cannot guarantee.
Bridges are the bottleneck. For a cross-L2 arbitrage, assets must traverse canonical bridges like Arbitrum's L1 Gateway or Optimism's Standard Bridge. The settlement latency on these bridges (often 1-7 days for full withdrawal) creates a massive window for MEV extraction and breaks atomicity. A shared sequencer does not accelerate L1 finality.
Intent-based solvers win. Protocols like UniswapX and Across already abstract this complexity. They use a network of solvers to compete for cross-domain bundles, routing users via the optimal path (e.g., Hop, Stargate). A shared sequencer cannot match the economic efficiency of this permissionless solver market for cross-chain flow.
Evidence: LayerZero's dominance. The dominant cross-chain messaging layer, LayerZero, processes orders of magnitude more cross-chain transactions than any shared sequencer handles. This proves the market demand is for generalized, application-layer interoperability, not a monolithic sequencing layer attempting to force atomicity.
takeaways
WHY SHARED SEQUENCING ISN'T THE FINAL LAYER
TL;DR for Architects
Shared sequencing solves for atomic composability within a rollup set, but cross-rollup MEV requires a separate, higher-order solution.
01
The Atomicity Illusion
Shared sequencers like Astria or Espresso guarantee atomic bundles within their domain (e.g., all rollups on Celestia). Cross-domain execution (Ethereum <> Arbitrum <> zkSync) is still asynchronous, leaving $100M+ in cross-chain arbitrage MEV on the table.\n- Domain Boundary Problem: Atomicity ends at the sequencer's L1 data availability layer.\n- Latency Arbitrage: The ~12s Ethereum block time creates a massive window for searchers.
~12s
Arbitrage Window
$100M+
Annual MEV
02
The Settlement Layer Bottleneck
Even with a shared sequencer, finality requires settlement on an L1 (e.g., Ethereum). This creates a two-phase commit where intent is signaled in the sequencer layer but finalized later, allowing for time-bandit attacks. Projects like Across and Chainlink CCIP solve this with optimistic or cryptographic verification, not sequencing.\n- Finality vs. Ordering: Shared sequencing provides ordering, not finality.\n- Verification Overhead: Secure bridging requires its own fraud/validity proof system.
2-Phase
Commit Process
30min+
Challenge Periods
03
Intent-Based Architectures Win
The endgame is intent-based cross-domain systems like UniswapX, CowSwap, and Across, which abstract settlement logic away from users. They use fillers and solvers to compete on execution, internalizing MEV as better prices. Shared sequencing is a component, not the architecture.\n- Solution vs. Protocol: Shared sequencing is infra; intent protocols are application-layer solutions.\n- MEV Internalization: Searcher competition becomes user surplus.
90%+
Fill Rate
0 Gas
User Experience
04
The Sovereignty Tax
Rollups adopt shared sequencers for scale but cede sovereignty over their state transition function. This creates a single point of liveness failure and complicates forced inclusion for censorship resistance. A truly decentralized sequencer set (like EigenLayer) faces the latency/security trilemma.\n- Centralization Vector: Few sequencer operators control the order flow for many chains.\n- Censorship Resistance: Requires fallback to a slower, decentralized L1.
1
Liveness SPOF
~100ms
Latency Floor
ENQUIRY
Get In Touch
today.
Our experts will offer a free quote and a 30min call to discuss your project.
NDA Protected
Confidentiality24h Response
Time to ValueDirectly to Engineering Team
No Sales Fluff10+
Protocols Shipped
$20M+
TVL Overall