Blockchain

The L2 Consolidation Is Here: Arbitrum and Base Have Already Won

By Anurag Verma•June 4, 2026

The L2 Consolidation Is Here: Arbitrum and Base Have Already Won (And Most Rollups Know It)

Seventy-three active rollups now secure more than $48 billion in total value locked across the Ethereum Layer 2 landscape as of mid-2026. But that headline number is deceptive. Strip it apart and you find two chains, Arbitrum One and Base, holding roughly 77% of all L2 DeFi liquidity between them. The war is mostly over. What we are watching now is the cleanup.

I have been tracking rollup infrastructure since the early Optimism days, and I do not think most people have fully processed how decisive this consolidation has become. Arbitrum One alone sits at approximately $16.9 billion TVL, capturing somewhere between 40 and 44 percent of the entire L2 market. Base follows at around $12.8 billion, a number that looked impossible in 2023 when Coinbase launched the chain. The remaining $18 billion is distributed across 71 other rollups, most of which are fighting for scraps.

What Actually Decided the Winner

The naive theory was that fees would determine who won. Cheaper data posting equals more users. That logic had merit in 2022 and early 2023, but EIP-4844 (Dencun) invalidated it for almost everyone by early 2024. Blob transactions collapsed L2 data costs by 90 to 99 percent overnight. Once the floor dropped that low, fee differentiation ceased to matter at the margins that regular users notice. You cannot win a race by being slightly cheaper when everyone is already at fractions of a cent.

The Pectra upgrade in May 2025 doubled down on this, adding EIP-7691 to further expand blob throughput for L2s. Then Fusaka landed in December 2025, introducing PeerDAS and scheduling blob capacity increases from 6 to eventually 14 per block in staged forks. L2 fees are projected to fall another 40 to 60 percent as these changes roll out. My view: this is great for users and brutal for any chain whose pitch was "we are fast and cheap." That pitch is now table stakes, not differentiation.

What actually decided the consolidation was distribution and depth. Coinbase had tens of millions of verified users and a mobile app. They plugged Base into that funnel directly. The result: Base became the most active L2 by transaction count, generating roughly 46 percent of all L2 transactions and reportedly the only major L2 to turn a profit in 2025, earning around $55 million. That is not a scaling story. That is a distribution story.

Arbitrum won on institutional depth. Stablecoin balances crossed $5 billion in October 2025. Then Robinhood announced its own Arbitrum Orbit L2, processing 4 million transactions during its first testnet week in February 2026. Financial services firms are choosing Arbitrum because the liquidity is already there, because GMX and other DeFi protocols built there first, and because the developer tooling matured before the competitors caught up.

The OP Stack Play Is Different

Optimism's market position is harder to read if you only look at TVL for OP Mainnet itself, which sits around $1.9 billion. That number misses the point. Optimism bet on the Superchain model: a shared codebase where every chain that builds on the OP Stack contributes sequencer revenue back to the collective and eventually shares liquidity natively. The Superchain now comprises 30-plus chains handling over 7 million daily transactions, with nearly half of all Ethereum L2 transactions happening on OP Stack chains collectively. Kraken's Ink launched in late 2024 as a DeFi-focused Superchain member. Interoperability between Superchain members, letting users move assets without external bridges, is rolling out now.

I think this is the most technically interesting bet in L2 land right now. Whether it pays off at the chain-level depends on whether shared sequencing actually ships without breaking things.

ZK Rollups: Still Technically Right, Still Losing the User Race

The ZK camp is consolidating too, just more painfully. zkSync Lite shut down on May 4, 2026, with Matter Labs freezing the network's final state so users could still exit while all development shifted to zkSync Era and the ZK Stack. By the time it closed, fewer than 200 transactions per day were passing through it. The first ZK rollup, retired with a whimper.

Starknet remains around $617 million TVL, drawing developers who want to build Cairo-native applications rather than porting EVM code. There is a real case for Starknet in gaming and applications that need ZK proofs in the compute layer, not just the settlement layer. But Starknet has never cracked mainstream DeFi liquidity, and nothing in the current trajectory suggests it will before Arbitrum and Base cement their leads further.

The uncomfortable truth for ZK maximalists is that "cryptographically superior" does not win markets. Better developer experience and distribution do. ZK proofs add latency and complexity compared to optimistic fraud proofs, and the period when that trade-off might have attracted users willing to tolerate friction has passed.

The 73-Chain Problem

21Shares research noted that most Ethereum L2s may not survive 2026 as activity concentrates. Blast's TVL collapsed 97 percent from $2.2 billion in June 2024 to around $55 million by December 2025. Scroll, Linea, and a dozen others are clinging to niche positions. The yellow.com L2 analysis from Q1 2026 described this as a split between winners and dead weight, and I think that framing is accurate.

The remaining differentiation vectors are narrow: application-specific rollups (Robinhood Chain, Ink, and similar exchange chains) that do not need to win open market TVL because they have captive users; and ZK-native chains targeting specific compute use cases. General-purpose "we are also a rollup" chains without distribution anchors are in trouble.

What Comes Next

The Glamsterdam upgrade is targeted for late 2026, aiming to improve the operational load capacity that PeerDAS opened up. If the blob expansion roadmap holds, Ethereum L2 fees should hit a floor that makes most current fee comparisons irrelevant.

By Q3 2026, L2 TVL is projected to exceed Ethereum L1 DeFi TVL for the first time. That milestone will get celebrated. But I will be paying more attention to sequencer revenue concentration and whether any challenger can pull liquidity from Arbitrum's DeFi base or match Coinbase's distribution for Base. So far, nothing on the horizon looks capable of doing either in the near term.

The rollup wars produced exactly two dominant general-purpose chains, a federation of OP Stack chains, and a long tail that will mostly consolidate or disappear. That is a normal market outcome. It is just not the permissionless pluralism the early rollup narrative promised.