The L2 Consolidation Moment: Two Chains Own 87% of the Market While One ZK Darling Faded

The narrative around Ethereum L2s in 2024 was all about the ZK vs Optimistic debate — which proof system would win, which chain would absorb the most liquidity. As of mid-2026, that debate has a fairly clear answer in terms of TVS rankings, and it's not what the ZK maximalists expected.

Numbers from L2Beat as of May 23, 2026.

## The Actual Rankings

Arbitrum One holds $15.75B in total value secured. Base sits at $12.14B. Those two together account for roughly 87% of the top-20 L2 TVS. OP Mainnet trails at $1.52B. Then Mantle at $1.46B, which is actually a ZK-validity rollup using SP1 Hypercube — a small but interesting counter-data point.

ZKsync Era is at $294M. For context: it was once positioned as *the* flagship ZK rollup, the chain that would eat Arbitrum's lunch through technical superiority. It's now running 0.19 user operations per second — the same throughput as Scroll, which holds $53M in TVS. The activity gap between ZKsync Era and Base is approximately 700:1.

## Base vs Arbitrum: Different Wins

The interesting thing is that these two chains have *different* stories even at similar TVS levels.

Arbitrum: $15.75B in TVS, 14.00 UOPS. Deep DeFi liquidity. The chain where large protocols park capital.

Base: $12.14B in TVS, 132.94 UOPS. 9.5x more active than Arbitrum by user operations. $2.19K/day in Ethereum data costs for 1.54 GiB of daily blob data — borderline absurd efficiency post-Pectra.

Then in March 2026, Base formally decoupled from the Optimism Superchain. They share the OP Stack codebase with OP Mainnet, but now have independent governance and their own upgrade path. Whatever Coinbase's long-term play is here, they've signaled they don't want to be subordinate to Optimism's governance structure.

## What Fusaka Changed

Ethereum's Fusaka upgrade went live in December 2025, and PeerDAS is the relevant piece. Each node now stores only 1/8th of blob data; any 50% of the network can reconstruct 100%. The practical result is roughly 8x DA capacity increase, with blob-parameter-only forks that let Ethereum raise blob limits *between* major upgrades without a full hard fork.

That last part is more significant than it sounds. L2 scaling is now partially decoupled from Ethereum's major upgrade schedule. The infrastructure question has an answer, at least for the next few years.

## ZKsync Stage 0 Problem

The least-discussed part of ZKsync Era's decline: it's still at Stage 0 in L2Beat's maturity classification. Most top-10 L2s are at Stage 1 (Arbitrum, Base, OP Mainnet, Mantle, Ink). Stage 0 means the highest centralization risk — the operators can theoretically override user funds without cryptographic proof constraints holding them back.

A chain that led the ZK space by narrative but ended up both losing users and sitting at Stage 0 is a strange outcome. ZKsync's V29 interop upgrade via the ZKsync Gateway launched in October 2025, aiming for native multi-chain messaging — hasn't moved the activity needle.

## The Consolidation Pattern

Arbitrum and Base winning isn't really about proof systems. Optimistic rollups won because they had working products, liquidity, and developer ecosystems first. The ZK proof advantage that everyone expected to matter by now hasn't translated into user preference in practice.

There are now 6+ ZK rollups in the top 20 by TVS (Mantle, Starknet, Linea, ZKsync, Scroll, Katana). Together they're a fraction of Arbitrum + Base. Starknet at $621M with its Stwo prover is the most interesting ZK chain at this point — if any of them eventually close the gap, that's the one to watch.

The 2026 L2 story is: infrastructure problems are largely solved, consolidation happened faster than expected, and the winning chains won on execution and distribution, not just on proving system elegance.

The L2 Consolidation Moment: Two Chains Own 87% of the Market While One ZK Darling Faded

// COMMENTS

ON THIS PAGE