The Ethereum Layer 2 ecosystem entered 2026 with more than forty live rollup networks and a growing consensus that this number is unsustainable. The fragmentation of liquidity, user attention, and developer effort across dozens of competing chains has produced an environment where most L2s are running at low utilization, subsidizing transactions to attract users, and competing for a share of Ethereum’s blockspace that is simultaneously abundant and contested. Consolidation is not a prediction — it is already underway.

The Rollup Landscape by the Numbers

Arbitrum One holds the dominant position in total value locked, with approximately $18 billion at the start of Q1 2026. Base, Coinbase’s L2, has grown rapidly on the back of its parent company’s distribution, reaching roughly $12 billion TVL. Optimism’s OP Mainnet, despite being the architectural progenitor of both Base and several other OP Stack chains, trails both in TVL but leads in developer ecosystem contributions through the Superchain initiative.

zkSync Era and StarkNet represent the ZK-rollup category — cryptographically more elegant, computationally more expensive, and technically more complex to build on. ZK-rollups carry a theoretical security advantage: validity proofs guarantee correct execution without requiring the challenge period that optimistic rollups depend on. In practice, the 7-day withdrawal period that defines optimistic rollup risk has not materialized as a user-experience crisis because liquidity bridges absorb the wait time for most users.

Polygon’s zkEVM and Scroll complete the ZK picture. Neither has achieved the TVL or transaction volume of the leading optimistic rollups, but both have meaningful enterprise and institutional pipeline, particularly in Southeast Asia and the Middle East.

Dencun Changed the Economics

Ethereum’s Dencun upgrade in March 2024 introduced EIP-4844, which created a new data availability layer — “blobs” — for rollups to post transaction data to Ethereum at dramatically lower cost. The effect on L2 economics was immediate and large. Transaction fees on major rollups dropped by 80-90% within weeks of the upgrade. Arbitrum’s average transaction fee fell from approximately $0.40 to under $0.05. Base’s fees dropped below $0.01 for simple transfers.

This was good for users. It was complicated for L2 business models. Most rollup revenue derives from the spread between the fee charged to users and the cost of posting data to Ethereum — the “sequencer spread.” Dencun compressed that spread significantly. L2s that were marginally profitable became marginally unprofitable. The pressure to achieve scale intensified.

The Superchain and the Aggregation Play

The OP Stack’s Superchain concept — a network of interoperable L2 chains sharing a common standard and message-passing infrastructure — is the most coherent answer to the fragmentation problem. Base, OP Mainnet, Mode, Zora, and a growing list of institutional chains all run on the OP Stack. They share infrastructure, contribute sequencer revenue to the Optimism Collective’s retroactive funding mechanism, and are building toward native cross-chain interoperability.

The aggregation logic is sound: if you can’t have one chain dominate, create a standard that makes the chains interoperable enough to function as one from the user’s perspective. Liquidity stays separated, but the friction of moving between chains decreases. The Superchain does not solve the fragmentation problem entirely — it solves it within the OP Stack ecosystem while remaining fragmented relative to Arbitrum and the ZK-rollup ecosystem.

Which Chains Survive

The consolidation will be driven by two factors: distribution and utility. Distribution means an L2 attached to a company or institution with an existing user base — Coinbase’s Base is the canonical example. Utility means an L2 that has captured a specific use case deeply enough that users would not move — Immutable’s gaming-focused chain, or the various DeFi-specific L2s that have become dominant venues for specific protocols.

Generic, undifferentiated L2s with no distribution advantage and no dominant use case are already seeing flatlined or declining metrics. They will not close — the code continues to run — but their sequencer revenue will approach zero as users migrate to chains with better liquidity and lower fees. The “40 L2s” headline will not become “5 L2s” through formal closures. It will become “5 L2s that matter” through market gravity, which is a quieter but equally decisive outcome.