What shared sequencing actually is
Every Layer 2 rollup needs a sequencer: the component that receives transactions, picks their order, and produces blocks before they settle to Ethereum. Almost all of today's major rollups run a single, centralized sequencer operated by the rollup team. It works, but it's a trust assumption and a liveness chokepoint, and it leaves each rollup as an isolated island.
Shared sequencing replaces those siloed sequencers with one network that orders transactions for many rollups at once. Rollups submit their transactions to a shared layer, which runs a consensus protocol to produce a canonical ordering, then hands ordered batches back to each rollup's execution and proving pipeline. The rollups keep their own state and proofs — only the ordering job is pooled.
How it works
A shared sequencer typically separates ordering from execution. The sequencing network reaches consensus on a sequence of transactions spanning multiple rollups; each rollup's nodes then execute only the slice relevant to them. Because all the rollups in the set share one ordering authority, you can express a transaction that is valid only if a matching transaction on another rollup also lands in the same block — the basis for atomic cross-rollup composability. That's the headline use case: swap on rollup A and supply on rollup B as one indivisible unit, with no bridge-and-wait gap for arbitrage to exploit.
Espresso is the most prominent live design. Its HotShot consensus targets fast finality — roughly six-second confirmation versus Ethereum's multi-minute window — and its CIRC protocol aims at both asynchronous and synchronous cross-rollup calls. A competing school of thought is based sequencing, where ordering is handed back to Ethereum L1 validators directly. Taiko brands itself as the first based rollup, leaning on L1 proposers plus preconfirmations — restaked economic promises that a transaction will be included — to hide L1's 12-second latency. "Based Espresso" blurs the two by offering ad-hoc shared sequencing to based rollups and validiums alike.
Why it matters
The modular thesis broke Ethereum scaling into execution, data availability, and settlement. Sequencing was the leftover piece, and it's where two real problems live: rollups can't talk to each other synchronously, and centralized sequencers are a single point of censorship and failure. Shared sequencing attacks both. It also reframes restaking as a security primitive — preconfirmation guarantees are increasingly backed by restaked ETH, so a sequencer that reneges gets slashed.
Key risks and tradeoffs
The uncomfortable truth: a shared sequencer that orders transactions across many rollups is also the single richest venue for cross-domain MEV. You don't eliminate the chokepoint — you consolidate it. In my view that's the unsolved problem of the category, and it's why encrypted-mempool approaches (Radius uses verifiable delay encryption to hide transaction content until ordering is fixed) keep showing up alongside shared sequencers rather than as afterthoughts.
There's also an adoption-economics problem. A shared sequencer is only valuable if enough high-liquidity rollups join the same set; otherwise you've added a dependency for composability nobody uses. That gap is what sank Astria, the Celestia-based shared sequencer that halted its network in early December 2025 at block 15,360,577 after roughly $12.5M in funding and thin adoption. The technology worked; the two-sided market didn't materialize. For LCX-style regulated venues, there's a further wrinkle: outsourcing ordering to a third-party network raises real questions about who is responsible for fair-ordering and market-integrity obligations.
Current state (2026)
The field has consolidated faster than most expected. Astria's shutdown removed one of the two flagship general-purpose networks, leaving Espresso as the main live shared-sequencer bet while based sequencing absorbs much of the remaining mindshare because it inherits Ethereum's neutrality rather than building a new trust layer. The data suggests rollup operators are hedging — running on more than one sequencing market rather than committing — which tells you the question of who orders blockchains is still genuinely open.
Watch whether a major existing rollup with real liquidity actually migrates its ordering to a shared layer in production. Until one does, shared sequencing remains a strong idea waiting for its first irreversible commitment.