Shared Sequencers & Decentralized Sequencing for L2 Rollups
Centralized sequencers are one of the biggest unsolved problems in rollup infrastructure. Every major L2 (Arbitrum, Optimism, Base, zkSync, Starknet) currently relies on a single sequencer entity that can censor transactions, extract MEV, and create single points of failure. Shared sequencers and decentralized sequencing approaches promise to solve this. This guide explores how Espresso, Radius, and based rollups are enabling censorship-resistant, composable L2 ecosystems.
1. The Sequencer Problem
In the rollup design, a sequencer is a critical infrastructure role: it collects pending transactions from users, orders them, and submits them to L1 as a batch. This ordering power is enormous. The sequencer decides transaction inclusion, priority, and ordering—which directly impacts MEV extraction, transaction censorship, and network liveness.
This is one of those topics where surface-level understanding is dangerous. We've seen traders lose significant capital from misconceptions covered in this guide.
Currently, all major rollups use centralized sequencers:
Arbitrum
Single Sequencer (Offchain Labs), can earn MEV and propose transaction ordering
Optimism & Base
Centralized sequencer run by Optimism Foundation, sole transaction ordering
zkSync
Centralized sequencer (Matter Labs), plans decentralization long-term
Starknet
Centralized sequencer (Starkware), committed to progressive decentralization
The risks are severe:
Centralized Sequencer Risks
Transaction Censorship: The sequencer can refuse to include your transaction, freezing assets or preventing liquidations in lending protocols.
MEV Extraction: The sequencer sees pending transactions and can sandwich trades, front-run liquidations, and reorder blocks to extract maximum value at user expense.
Liveness Failure: If the sequencer goes down, the entire rollup halts. There's no fallback ordering mechanism—the chain stalls until the sequencer returns.
Forced Withdrawals Delay: In bankruptcy scenarios, users can force withdraw from L1 after a delay, but this requires time and coordination.
This centralization contradicts the whole premise of blockchain: decentralization and censorship resistance. Solving sequencer decentralization is critical for mature L2 ecosystems.
3. Approaches to Decentralized Sequencing
There are three main approaches to solving the sequencer problem:
Shared Sequencer Networks
External decentralized networks (Espresso, Radius) that rollups opt into. The shared sequencer runs independently and serves multiple rollups. Rollups configure which shared sequencer to use and inherit its security model.
Based Sequencing (Based Rollups)
L1 Ethereum validators directly sequence L2 transactions. Instead of a separate shared sequencer network, the rollup uses Ethereum's own validator set. This is the ultimate in decentralization but trades latency for security. Taiko is the leading example.
Self-Sequencing (Decentralized Sequencer Sets)
Each rollup runs its own decentralized sequencer set (a network of validators managed by the rollup). Some rollups plan this long-term (Arbitrum, Optimism), but it's more complex to implement than shared or based sequencing and lacks cross-rollup benefits.
The tradeoff: Shared sequencers offer sub-second finality and cross-rollup composability but require trusting a separate network. Based rollups inherit L1 security but with higher latency. Self-sequencing gives each rollup control but lacks composability benefits.
4. Espresso Systems: The Leader
Espresso is the most advanced shared sequencer network and the clear market leader.
Funding & Momentum
Espresso raised $28M Series B led by a16z crypto in March 2024. The company transitioned to full PoS with permissionless participation and launched the ESP token in early 2026 with a 10% airdrop (total supply: 3.59B tokens). Mainnet confirmation layer went live in November 2024, and the network now has 20+ chain partnerships.
How Espresso Works
Espresso uses HotShot consensus (a fast PoS protocol) with economic security backed by EigenLayer restaking. Validators stake ESP (or liquid staking derivatives of ETH via EigenLayer) to participate. The network orders transactions from multiple rollups in a single shared sequence, then publishes commitments to Ethereum L1.
Key features:
Sub-second Finality: Espresso targets sub-second transaction finality, much faster than Ethereum's 12-second slots.
Cross-Rollup Composability: Atomic transactions can reference state from multiple rollups, enabling swaps and lending across L2s.
EigenLayer Security: Validators slashing conditions ensure honest sequencing; malicious sequencers lose staked capital.
MEV-Fair Ordering: Uses fair ordering protocols to minimize MEV extraction, though not eliminating it entirely.
Rollup Flexibility: Works with optimistic rollups, ZK rollups, and hybrid architectures.
Partners & Adoption
Espresso has partnerships with 20+ chains:
ApeChain
Ethereum L2 for the Ape Coin ecosystem
RARI Chain
Raydium's dedicated DEX rollup
Celo
Mobile-first L1 transitioning to L2
Cartesi & Polygon AggLayer
ZK rollups and rollup aggregation layers
This adoption rate suggests strong product-market fit. Rollups are willing to give up sequencer revenue for Espresso's security, composability, and MEV mitigation benefits.
5. Radius: Encrypted Sequencing
Radius takes a different technical approach: instead of a fair-ordering consensus protocol, Radius uses an encrypted mempool to prevent MEV entirely.
The Encryption Approach
Radius uses PVDE (Practical Verifiable Delay Encryption)—transactions are encrypted with a time-locked key so the sequencer can't see their contents. The sequencer orders encrypted bundles without knowing what's inside, making MEV extraction impossible. After a delay, the encryption decrypts and transactions execute.
Key differences from Espresso:
MEV Prevention (not mitigation): Radius actually prevents MEV, whereas Espresso mitigates it. The sequencer literally can't see transaction contents.
Simpler Consensus: No need for complex fair-ordering protocols; standard BFT consensus works fine since sequencers can't extract MEV anyway.
Latency Trade-off: The encryption decryption delay adds latency; finality is ~1 second.
Blockchain Agnostic: Works with any rollup type (optimistic, ZK, validium, etc.).
Funding & Status
Radius raised $1.7M in pre-seed funding led by Hashed (May 2024). The project is in testnet with partnerships including Saga (RaaS), Celestia ecosystem projects, and AltLayer integration. Mainnet launch is expected in 2026.
Radius appeals to projects prioritizing MEV resistance over extreme latency optimization. The encryption approach is cryptographically elegant but adds complexity around key management and threshold decryption.
6. Based Rollups: An Alternative
Based rollups take a fundamentally different approach: instead of external shared sequencers, use L1 validators to sequence L2 transactions. The rollup is built directly on Ethereum consensus.
How Based Rollups Work
L1 validators propose L2 transactions directly in Ethereum blocks (as calldata or blobs). The L2 decodes these transactions and executes them in block order. This means L2 inherits Ethereum's full validator set for sequencing—no separate network to trust.
Taiko: The Leading Example
Taiko is the most advanced based rollup. It uses a ZK-based validity proof and based sequencing (L1 proposers). Taiko has mainnet running on Ethereum with full based sequencing. The L2 inherits Ethereum's 800K+ validators, making censorship and shutdown impossible without attacking Ethereum itself.
Trade-offs
Based rollups have tradeoffs compared to shared sequencers:
Higher Latency: L2 finality is tied to L1 slot time (12 seconds). Shared sequencers offer sub-second ordering.
Preconfirmations: To address latency, sequencers can provide preconfirmations (signed promises of inclusion) that offer faster finality while L1 confirmation catches up.
Maximum Security: Inheriting L1 validator set means full Ethereum security—the strongest possible assumption.
Limited Composability: Cross-rollup composability is harder; each based rollup uses L1 for sequencing, so they can't directly communicate atomically.
Liveness Guarantees: Same as Ethereum—as long as Ethereum is alive, the based rollup works.
For maximum decentralization and security at the cost of latency, based rollups are the answer. For speed and composability, shared sequencers win. The best choice depends on use case.
7. Comparison of Approaches
Here's how the three approaches stack up:
| Feature | Centralized Sequencer | Shared Sequencer (Espresso) | Encrypted Sequencer (Radius) | Based Rollup |
|---|---|---|---|---|
| Decentralization | Single operator | PoS network (800+ validators planned) | Encrypted mempool | L1 validators (800K+) |
| Censorship Resistance | Low | High | Very High (encrypted) | Highest (L1 level) |
| MEV Protection | None | Partial (fair ordering) | Strong (encrypted txs) | Depends on L1 MEV |
| Cross-Rollup Composability | None | Yes (atomic bundles) | Yes (atomic bundles) | Limited |
| Finality | ~250ms | ~1 second | ~1 second | 12s (w/o preconfs) |
| Liveness Risk | Single point of failure | Distributed | Distributed | L1 liveness |
| Adoption Status | Production (all major L2s) | Mainnet live (20+ chains) | Testnet (2026 launch) | Early mainnet (Taiko) |
8. Risks & Challenges
Despite the promise of shared sequencers, significant hurdles remain:
Economic Barriers
Rollups earn substantial revenue from transaction fees that include sequencing profit. Switching to an external shared sequencer means sharing or losing this revenue. Arbitrum and Optimism alone earn billions annually from sequencing. This is the biggest adoption barrier—rollups are economically incentivized to keep sequencing centralized.
Astria's Shutdown: A Cautionary Tale
Astria was a Celestia-based shared sequencer that raised $5.5M from 1kx, Delphi Ventures, and Figment Capital. It launched mainnet in October 2024 but ceased operations on December 2, 2025. The failure highlights that even well-funded shared sequencing projects can struggle to achieve product-market fit if rollups don't actually adopt them.
Factors contributing to Astria's shutdown likely included:
Low Adoption: Few rollups actually adopted Astria; most preferred centralized sequencing or waited for more mature solutions.
Revenue Sharing Negotiations: Rollups were unwilling to commit to paying Astria a significant portion of sequencing revenue.
Competitive Pressure: Espresso moved faster to mainnet with stronger backing, making Astria's value proposition unclear.
Operational Challenges: Running a distributed sequencer network with multiple rollup clients is technically complex.
Regulatory Uncertainty
Shared sequencer networks introduce new regulatory questions. Are sequencers money transmitters? Do they have obligations around transaction ordering? How are they liable if they censor transactions? Clarity on these questions is still emerging.
New Trust Assumptions
While shared sequencers are better than single operators, they still introduce new trust assumptions. A cartel of Espresso validators could theoretically censor transactions or collude. Encrypted approaches (Radius) mitigate this better. Based rollups inherit L1 security entirely.
Latency Trade-offs
Shared sequencers can't match centralized sequencers' ~250ms latency due to network overhead and consensus delays. Based rollups trade even more latency for security (12+ seconds). Preconfirmations help but add complexity.
9. FAQ
What is a shared sequencer?
A shared sequencer is a decentralized network of sequencers that can serve multiple rollups simultaneously. Instead of each rollup running its own centralized sequencer, they outsource sequencing to a shared, decentralized network. This provides censorship resistance, MEV mitigation, cross-rollup composability, and liveness guarantees. Espresso and Radius are the leading examples.
Why do rollups need decentralized sequencing?
Centralized sequencers can censor transactions, extract MEV from users, and create single points of failure that halt the entire rollup. Decentralized sequencing provides censorship resistance, fairer transaction ordering, better liveness guarantees, and enables atomic cross-rollup composability. It addresses one of the biggest unsolved problems in the rollup ecosystem.
What happened to Astria?
Astria was a Celestia-based shared sequencer that launched mainnet in October 2024 but ceased operations on December 2, 2025. It highlighted the challenges of achieving product-market fit for shared sequencing infrastructure, as rollups are reluctant to give up sequencer revenue and adoption barriers remain high. Despite strong funding from 1kx and Delphi Ventures, lack of rollup adoption made the business unsustainable.
How does Espresso\'s shared sequencer work?
Espresso runs a PoS network using HotShot consensus, backed by EigenLayer restaking for economic security. Rollups opt into using Espresso for sequencing, gaining censorship resistance and cross-rollup composability. The ESP token launched in early 2026 with a 10% airdrop (3.59B total supply). Espresso has 20+ chain partnerships and targets sub-second finality. Validators stake ESP to participate and earn sequencing fees.
What are based rollups and how do they differ from shared sequencers?
Based rollups use L1 Ethereum validators to sequence L2 transactions, inheriting full decentralization. Shared sequencers are separate networks rollups opt into. Based rollups trade latency for maximum security (inherited from L1); shared sequencers offer faster confirmations but introduce new trust assumptions. Taiko is the leading based rollup example. Based rollups inherently have better security but worse latency than shared sequencers.
Will rollups actually adopt decentralized sequencing?
Adoption faces significant economic hurdles since rollups earn billions annually from sequencing revenue. However, competitive pressure, regulatory scrutiny around centralization, and user demand for censorship resistance are driving adoption. Espresso has 20+ chain partnerships, suggesting growing momentum. Rollups may adopt decentralized sequencing gradually, starting with lower-value chains and expanding as revenue-sharing models improve and network effects increase.
Related Guides
- Based Rollups: Ethereum L1 Sequencing Guide 2026 — Explore Taiko and other based rollups using L1 validators
- Ethereum Layer 2 Ecosystem Guide 2026 — Comprehensive overview of rollups, sidechains, and scaling solutions
- MEV Supply Chain & PBS Guide 2026 — Understand MEV extraction and proposer-builder separation
- Data Availability & Modular Blockchains Guide 2026 — How Celestia and DA layers complement shared sequencers
- Optimism Superchain & OP Stack Guide 2026 — Learn about Optimism\'s rollup ecosystem and cross-chain messaging
Educational Disclaimer
This guide is for educational purposes only and does not constitute financial, investment, or security advice. Shared sequencer infrastructure is rapidly evolving, and the landscape may change significantly. Always conduct your own research, understand the risks of using new infrastructure, and consult with security professionals before deploying capital or building on shared sequencers. The author and degen0x are not liable for losses or damages arising from decisions made based on this guide.
Educational disclaimer: This guide is for informational purposes only and does not constitute financial advice. Crypto involves significant risk — do your own research before making any decisions. Learn more about our team.
Educational disclaimer: This guide is for informational purposes only and does not constitute financial advice. Crypto involves significant risk — do your own research before making any decisions. Learn more about our team.