The L2 compromise is broken, it’s time for a new foundation

The second quarter of 2025 was a actuality verify for the scaling of the blockchain, and for capital preservation to pour in rollups and facet chains the cracks within the layer 2 mannequin grow to be bigger. The unique promise of L2s was easy: scale L1s, however the prices, delays, and fragmentation in liquidity and person expertise proceed to pile up.

New tasks like stablecoin funds are beginning to query the L2 paradigm, asking if L2s are actually safe, and are their sequencers extra like single factors of failure and censorship? Typically they’ll find yourself taking a pessimistic view that maybe fragmentation is just inevitable in web3.

Are we constructing a future on a strong basis or on a home of playing cards? L2s should face and reply these questions. If Ethereum’s (ETH) base consensus layer have been inherently quick, low cost, and infinitely scalable, the whole L2 ecosystem as we all know it might be redundant. Quite a few rollups and sidechains have been proposed as “L1s add-ons” to alleviate the elemental limitations of the underlying L1s. It is a type of technical debt, a fancy, fragmented answer that has been handed over to web3 customers and builders.

And to reply these questions it’s essential to deconstruct the whole idea of an L2 into its basic parts, to disclose a path to a extra strong and environment friendly design.

An anatomy of L2s

Construction determines operate. It’s a primary precept in biology that additionally applies to laptop methods. To find out the suitable construction and structure of L2s, we should fastidiously study their capabilities.

At its core, every L2 performs two essential capabilities: Sequencing, i.e. ordering transactions; in addition to calculating and proving the brand new state. A sequencer, whether or not a centralized entity or a decentralized community, collects, orders, and batches person transactions. This batch is then executed, leading to an up to date standing (e.g. new token balances). For safety causes, this standing should be settled on the L1 through Optimistic or ZK Rollups.

Optimistic rollups assume that each one standing transitions are legitimate and depend on a problem interval (typically seven days) throughout which anybody can submit proof of fraud. This creates a serious UX trade-off and lengthy finality instances. ZK Rollups use zero-knowledge proofs to mathematically confirm the correctness of every state transition earlier than it reaches L1, enabling near-instant finality. The drawback is that they’re computationally intensive and complicated to construct. ZK proofs themselves can comprise errors, which may result in catastrophic penalties, and formal verification of this, if in any respect possible, could be very costly.

Sequencing is a governance and design alternative for each L2. Some choose a centralized answer for effectivity (or maybe for censorship energy; who is aware of), whereas others choose a decentralized answer for better equity and robustness. Finally, L2s resolve how they need to do their very own sequencing.

Producing and verifying state claims is the place we will do a lot, a lot better when it comes to effectivity. As soon as a set of transactions has been sequenced, calculating the subsequent state is a purely computational process, and that may be carried out utilizing only a single supercomputer, targeted solely on uncooked pace, with out the overhead of decentralization. That supercomputer may even be shared between L2s!

As soon as this new state is claimed, its verification turns into a separate, parallel course of. An enormous community of verifiers can work in parallel to confirm the declare. That can be the philosophy behind Ethereum’s stateless shoppers and highly effective implementations, resembling MegaETH.

Parallel verification is infinitely scalable

Parallel verification is infinitely scalable. Regardless of how rapidly L2s (and that supercomputer) produce claims, the verification community can all the time catch up by including extra verifiers. The latency right here is exactly the verification time, a hard and fast, minimal quantity. That is the theoretical optimum if you happen to use decentralization successfully: to confirm, to not calculate.

After sequencing and standing verification, the L2’s job is sort of full. The ultimate step is to publish the verified standing to a decentralized community, the L1, for final settlement and safety.

This final step exposes the elephant within the room: blockchains are horrible settlement layers for L2s! The primary computing work is completed off-chain, however L2s must pay an enormous premium to finish an L1. They face a double overhead: the restricted processing capability of the L1, burdened by the overall, linear ordering of all transactions, creates congestion and excessive information posting prices. Moreover, they have to endure the L1’s inherent finality delay.

For ZK Rollups that is minutes. For Optimistic Rollups, this includes a weeklong problem interval, a mandatory however pricey safety trade-off.

Goodbye, the ‘whole order’ fantasy in web3

Ever since Bitcoin (BTC), individuals have carried out their finest to squeeze all of the transactions of a blockchain into one whole order. In spite of everything, we’re speaking about blockchains! Sadly, this “whole order” paradigm is a pricey fantasy and clearly exaggerated for L2 settlement. How ironic it’s that one of many world’s largest decentralized networks and the world’s laptop behaves identical to a single-threaded desktop!

It is time to transfer on. The long run is native, account-based ordering, the place solely transactions that work together with the identical account have to be ordered, permitting for top parallelism and true scalability.

International order clearly implies native order, however it is usually an extremely naive and simplistic answer. After 15 years of ‘blockchain’, it is time we open our eyes and create a greater future. The scientific area of distributed methods has already moved from the sturdy consistency idea of the Eighties (which blockchains implement) to the sturdy closing consistency mannequin of 2015 that unleashes parallelism and concurrency. Time for the web3 business to maneuver on as nicely, depart the previous behind and comply with future-oriented scientific progress.

The period of the L2 compromise is over. It is time to construct on a basis designed for the long run, from which the subsequent wave of web3 adoption will emerge.