Ethereum's rollup-centric roadmap was supposed to be the master plan: push execution to Layer 2s, keep Ethereum L1 as the secure settlement layer, and let a thousand rollups bloom. Three years after EIP-4844 slashed L2 data costs by 100x, the results are in — and they are not what anyone expected...
"If you create a 10,000 TPS EVM where its connection to L1 is mediated by a multisig bridge, then you are not scaling Ethereum." — Vitalik Buterin, Ethereum Co-Founder
Ethereum's rollup-centric roadmap was supposed to be the master plan: push execution to Layer 2s, keep Ethereum L1 as the secure settlement layer, and let a thousand rollups bloom. Three years after EIP-4844 slashed L2 data costs by 100x, the results are in — and they are not what anyone expected.
More than 50 Layer 2 networks now operate on Ethereum, collectively securing over $38 billion in bridged assets. But the ecosystem has consolidated with brutal efficiency: Base and Arbitrum alone capture 77% of all L2 DeFi TVL. Base processes over 60% of all L2 transactions. The other 47+ rollups are, in the words of 21Shares' latest research, unlikely to survive 2026. Meanwhile, mainnet fee revenue has collapsed 90% year-over-year as activity migrated to L2s, starving the EIP-1559 burn mechanism that was supposed to make ETH "ultrasound money." Ethereum's co-founder himself declared in February 2026 that the original L2 vision "no longer makes sense."
This report examines the economic dynamics of Ethereum's L2 consolidation: who is capturing value, who is losing it, and whether the Ethereum Foundation's interoperability roadmap can stitch the fragments back together before the ecosystem cannibalizes its own base layer.
The data tells an unambiguous story. Despite the proliferation of 50+ Ethereum rollups, a power-law distribution has taken hold with startling speed:
| Network | Share of L2 DeFi TVL | Estimated Annual Sequencer Revenue | Status | |---------|---------------------|-----------------------------------|--------| | Base | 46.6% | ~$93M | Profitable | | Arbitrum | 30.9% | ~$42M | Operating | | Optimism | ~12% | ~$26M | Operating | | All Others Combined | ~10.5% | Minimal | Most at a loss |
Base — Coinbase's Layer 2 built on the OP Stack — has emerged as the undisputed winner. It was the only L2 to turn a clear profit in 2025, earning approximately $55 million after Ethereum settlement costs. Its average daily revenue runs at approximately $185,000, with priority fees comprising 86% of revenue. As the sole sequencer operator, Coinbase captures this revenue stream with software-like margins, frequently exceeding 80%.
Arbitrum retains dominance in DeFi-native activity and gaming, while Optimism's Superchain strategy has attracted enterprise deployments. Together, these three networks process approximately 90% of all L2 transactions. The remaining 47+ chains split the crumbs.
This concentration mirrors every platform market in technology history. But there is a critical difference: the value being concentrated is not flowing back to the underlying platform.
The economic contradiction at the heart of Ethereum's scaling story can be stated simply: L2s have succeeded in scaling Ethereum's execution capacity while failing to scale Ethereum's economic value.
The numbers are damning. In 2025, total Layer 2 revenue dropped 53% to approximately $129 million, primarily due to lower end-user fees — a success for users, but a crisis for the economic model. Payments from L2s to Ethereum mainnet fell even more sharply, declining to roughly $10 million — less than 10% of total L2 revenue. Year-over-year, the fee volume routed from L2s to Ethereum mainnet has crashed 90%.
This collapse directly undermines the deflationary thesis. EIP-1559's burn mechanism depends on mainnet gas fees. With activity migrating to L2s that settle via cheap blobs rather than expensive calldata, the burn rate has cratered. Ethereum's annual supply growth now sits at 0.23% — not inflationary in a dramatic sense, but a far cry from the "ultrasound money" narrative that peaked when mainnet was burning more ETH than it issued.
The sequencer economics are particularly revealing. Base's sequencer — operated exclusively by Coinbase — collects approximately $120 million in total fees, of which only $10 million flows to Ethereum L1 for data availability and settlement. The remaining $110 million stays with Coinbase. This is not a bug; it is the architectural reality of centralized sequencing. The sequencer orders transactions, captures priority fees and MEV, and pays Ethereum only the minimum required for security guarantees.
The result: Ethereum scales successfully, but Coinbase, Offchain Labs (Arbitrum), and the Optimism Foundation capture the economic upside. ETH holders get security demand and a trickle of blob fees. As of March 20, 2026, ETH trades at approximately $2,327 — having underperformed Bitcoin consistently since 2024, even as Ethereum network activity hits record highs.
The consolidation has produced a growing cemetery of "zombie chains" — rollups that launched with fanfare, attracted liquidity through incentive programs, and then watched their ecosystems evaporate.
Blast is the canonical cautionary tale. Its TVL peaked at $2.2 billion in June 2024, fueled by airdrop speculation. By December 2025, it had collapsed 97% to approximately $55 million. The founder went silent. The official X account stopped posting in May 2025. Users migrated en masse to Base and Arbitrum.
Blast is not alone. Usage across smaller rollups dropped 61% as incentive cycles ended. Even major DeFi protocols have responded: Aave and Synthetix scaled back deployments on weaker L2s, citing poor liquidity and limited returns. The message from capital allocators is clear — deploying liquidity across 50 chains is economically irrational when three chains hold 90% of the users.
The zombie chain phenomenon also creates a compounding fragmentation tax on the ecosystem. Developers must maintain deployments across multiple chains. Users face confusing bridge experiences. Liquidity is split across dozens of pools rather than concentrated where it generates the most efficient markets. Research from late 2025 on automated market makers across Ethereum and rollups confirmed that this fragmentation raises borrowing costs and reduces trading efficiency.
On February 3, 2026, Vitalik Buterin published what may be the most consequential post of his career since the Ethereum whitepaper itself. He declared that the original rollup-centric roadmap — the idea that L2s would be the primary scaling mechanism for Ethereum — "no longer makes sense."
His reasoning was twofold. First, L2 progress toward Stage 2 decentralization (where rollups operate with full training wheels removed, governed only by smart contract logic and fraud/validity proofs) has been "far slower and more difficult than originally expected." As of March 2026, nearly all major L2s still run centralized sequencers. Only Taiko (a based rollup using L1 sequencing) and Metis (a PoS sequencer pool since March 2024) have achieved meaningful sequencer decentralization.
Second, Ethereum's own L1 is now scaling directly. The Glamsterdam upgrade roadmap targets significant gas limit increases, and the EIL interoperability framework promises to reduce cross-L2 friction. Buterin argued that L2s should be reconceived — not as Ethereum's primary scaling solution, but as a "spectrum of networks" offering differentiated value: privacy features, application-specific execution environments, ultra-fast confirmations, or non-financial use cases.
The crypto community's response was polarized. L2 teams saw it as a betrayal of the roadmap they had spent years building toward. Pragmatists called it overdue recognition of market reality. But the subtext was unmistakable: the era of "every project launches its own L2" is over.
The Ethereum Foundation's response to fragmentation is the Ethereum Interoperability Layer (EIL) — a trustless messaging system designed to make 55+ L2 rollups feel like a single chain.
EIL moves interoperability logic on-chain and into user wallets, eliminating dependence on third-party bridge operators (historically, crypto's most exploited attack surface — over $2.8 billion stolen from bridges since 2022). Users would send tokens, mint NFTs, and trade across different rollups without switching networks or using external bridges.
The rollout follows three phases:
EIL entered testnet in November 2025 and targets mainnet deployment in 2026. Wallets will need to integrate the EIL SDK or support ERC-5792, while existing ERC-4337 smart wallets can add a multichain validation module.
The ambition is staggering: aggregate $42 billion in bridged liquidity into a unified experience. But the timing is precarious. If most L2s die before EIL reaches mainnet, interoperability between three surviving chains is a very different engineering problem than stitching together 55. The question is whether EIL is a solution to fragmentation or a monument to a fragmentation that already resolved itself through market consolidation.
Amid the consolidation carnage, one architectural pattern has emerged as a potential resolution to the sequencer centralization problem: based rollups.
A based rollup delegates transaction sequencing to Ethereum's own L1 validators rather than running a proprietary sequencer. This eliminates the risk of sequencer censorship, reorging, or value extraction — the rollup inherits Ethereum's full validator set for transaction ordering.
Taiko is the leading implementation. In early 2026, it shipped preconfirmations live on Ethereum mainnet — the first production based rollup where sequencing is performed by Ethereum proposers. The economic alignment is fundamentally different: instead of value flowing to a centralized sequencer operator, it flows to Ethereum validators, strengthening the base layer's economic security.
The catch is performance. Based rollups are currently limited by Ethereum's own block times and throughput. They cannot offer the sub-second confirmations that centralized sequencers like Base's provide. For high-frequency DeFi and consumer applications, this tradeoff may be unacceptable. But for applications that prioritize censorship resistance and credible neutrality over raw speed, based rollups represent the closest alignment between L2 scaling and L1 value capture.
Several major L2 teams — including leaders at Base, Optimism, Arbitrum, and Scroll — have expressed support for incorporating based rollup principles, signaling that hybrid architectures may emerge: centralized sequencing for speed, with based fallback for censorship resistance.
Power-law consolidation is complete. Base (46.6% of L2 DeFi TVL), Arbitrum (30.9%), and Optimism (~12%) now dominate. Most of the remaining 47+ rollups are zombie chains unlikely to survive 2026.
The value extraction problem is structural, not cyclical. L2 sequencers captured ~$161 million in 2025 revenue while routing only ~$10 million to Ethereum L1 — a 90% YoY collapse in mainnet fee revenue that undermines ETH's deflationary thesis.
Vitalik's February 2026 reversal is a watershed. The acknowledgment that the original L2-centric roadmap "no longer makes sense" opens the door to native rollups, L1 scaling, and a fundamental reconsideration of Ethereum's economic architecture.
EIL is ambitious but may arrive too late. The Ethereum Interoperability Layer targets a unified 55-chain experience, but market consolidation may render most of those chains irrelevant before mainnet launch.
Based rollups are the economic alignment play. By returning sequencing to L1 validators, based rollups like Taiko offer a path where L2 scaling strengthens rather than parasitizes the base layer — but face real performance tradeoffs.
ETH's investment thesis is under pressure. Trading at ~$2,327 with a 0.23% annual supply growth, ETH has underperformed BTC as the market prices in the value-capture disconnect between network activity and token economics.
Ethereum's Layer 2 strategy has achieved exactly what it set out to do: massively scale execution capacity at low cost. By that measure, it is a technical triumph. Base processes millions of transactions daily at sub-cent fees. Total L2 value secured exceeds $38 billion. Activity is at record highs.
But the economic architecture has produced an unintended outcome: the value created by this scaling accrues to sequencer operators — primarily Coinbase, Offchain Labs, and the Optimism Foundation — rather than to Ethereum itself. The base layer has become what critics call "infrastructure with capped valuation," providing security guarantees while others capture the economic upside.
The Ethereum Foundation's response — EIL, based rollups, L1 gas limit increases, the Glamsterdam upgrade — represents a course correction. Whether it arrives in time depends on a race between engineering timelines and market forces that have already concentrated 90% of L2 activity into three chains and sent dozens of others to the graveyard.
For investors and builders, the implications are stark: the L2 proliferation era is over. The consolidation era has arrived. And the next chapter of Ethereum's story will be defined not by how many chains it spawns, but by whether the value they create ever flows back home.