The Ethereum ecosystem is undergoing a recalibration of its scaling thesis. The rollup-centric roadmap, formalized in 2020, operated under the assumption that Layer 1 (L1) would act as a court of last resort, while Layer 2 (L2) solutions would absorb the entirety of the transactional surplus. On-chain activity data and core protocol developments throughout 2025 and 2026 partially invalidate this assumption. The issue does not reside in the efficiency of rollups for compressing data, but rather in the market’s inability to value decentralized security and the accelerated scaling of native L1 throughput.
The most relevant indicator of this dysfunction is the divergence between User Operations per Second (UOPS) and Total Value Locked (TVL). Rollups currently process over 3,400 operations per second, surpassing the mainnet in volume. However, the aggregate TVL of the L2 ecosystem has declined to figures near $50 billion in the third quarter of 2026, representing a year-over-year contraction exceeding 13%.
This inverse correlation suggests that rollups are being utilized as low-friction execution channels, but not as high-value settlement layers. The opportunity cost of maintaining liquidity in an L2 environment, subject to bridges and multi-signature security mechanisms, exceeds the benefit derived from gas reduction.
Centralization as an Economic Byproduct
The decentralization filter imposed by Vitalik Buterin through the stages (Stage 0, 1, and 2) has exposed an operational reality. Most L2 chains remain at Stage 0 or Stage 1, relying on multi-signature security councils and centralized sequencers for transaction ordering.
The resistance to migrating to Stage 2 is not exclusively attributable to technical limitations in generating fraud or validity proofs. A structural economic incentive exists: retaining control over the sequencer allows operators to capture Sequencer Extractable Value (SEV) and manage liquidity without exposure to L1 finality.
Maintaining a system at Stage 0 or 1 reduces infrastructure operating costs and offers flexibility to upgrade code without adhering to L1 governance timelines. This business logic collides with the original narrative of the “rollup as an extension of Ethereum.” If a rollup retains the ability to freeze or modify state via a private key, its security model more closely resembles that of a custodial sidechain than a trustless verification layer.
The Reduction of Opportunity Cost on L1
Progress in Layer 1 scalability alters the cost-benefit ratio of L2s. The implementation of EIP-4844 (Blobs) drastically reduced data publication costs, but the subsequent proposal to raise the gas limit to 80 million, and the projected target of 10,000 TPS for L1 under Justin Drake’s Strawmap roadmap, modify the competitive perimeter.
An increase in L1 capacity reduces the congestion price of block space. When L1 gas drops below certain thresholds, the comparative advantage of the rollup in terms of cost savings diminishes. The scaling equation cannot be based solely on data compression if the base layer increases its native bandwidth.
This phenomenon causes general-purpose rollups to face a commoditization of their service: by not offering substantial differences in EVM execution, they compete exclusively on price, which erodes margins and disincentivizes investment in zero-knowledge proof research.
The Shift Toward Native Rollup (EIP-8079)
The proposal for a Native Rollup, materialized in part through the discussion of the EXECUTE precompile (EIP-8079), proposes a reassignment of the validation role. Instead of the rollup sending a state and awaiting external verification, L1 would execute a re-execution of the L2 block via this precompile.
This design implies two technical consequences:
-
Verification Standardization: The validity of the L2 block will depend on the Ethereum execution client, not on a project-specific smart contract. This eliminates the need for each rollup to implement its own proof system, delegating responsibility to the base protocol.
-
Reduced Bridge Dependency: By being verified directly by L1 consensus, asset transfers between L1 and L2 are simplified, eliminating the 7-day challenge periods characteristic of Optimistic Rollups.
However, the adoption of Native Rollup presents an entry barrier. It forces teams to align their upgrade cycles with Ethereum hard forks, eliminating the flexibility to implement emergency patches without passing through L1 governance. For projects operating in high-risk sectors (complex DeFi or gaming), this rigidity represents an operational risk that many are unwilling to assume.
Functional Specialization vs. Horizontal Scaling
Vitalik Buterin has suggested a paradigm shift: L2s should cease to be viewed as “branded shards” of Ethereum and become specialized execution environments. This implies abandoning the competition for transactional volume and focusing on verticals that L1 cannot optimize by itself:
-
Privacy: Implementation of ZK proofs to hide sender or amount, something the public L1 cannot offer.
-
Specific Application Logic: Game engines with sub-millisecond block times or limit-order exchanges outside the AMM environment.
-
Alternative VM Languages: Environments that are not EVM (WASM, Move) but that settle assets on Ethereum L1.
This repositioning implies accepting that the ecosystem will not scale as a single global state machine, but as a heterogeneous set of execution layers with varying levels of trust. Interoperability will cease to be an automatic process and will become a problem of routing and state verification, where the end-user must assess the risk of each hop between layers.
The scaling model based exclusively on rollups has reached a point of diminishing returns. Data demonstrates that economic activity on L2 does not translate into proportional value capture for mainnet security, and the sluggishness in sequencer decentralization introduces counterparty risk that the original “trustless” narrative cannot sustain.
The industry faces a strategic fork: either accept that rollups are, in essence, centralized databases with lazy verification, or adopt the Native Rollup infrastructure to homogenize security, sacrificing iteration speed. The third path, proposed by Buterin, involves functional differentiation, where the rollup ceases to be a substitute for L1 and becomes an accelerator for specific use cases.
The market has already signaled its preference: liquidity takes refuge in the few rollups with the highest market share (Base, Arbitrum, Optimism), while the rest of the ecosystem faces a “subsidy trap” that it cannot sustain in the long term. The technical evolution toward Native Rollup is inevitable if the premise of security inheritance is to be preserved, but its implementation will clash with the commercial interests of current operators.
The sector must assume that Ethereum scaling is not an engineering problem of data compression, but a game theory problem between L1 and its operators. Decentralization is not a binary attribute achieved by default, but an operational cost that many participants have decided not to assume. The future of the network will depend on the base protocol’s ability to impose verification rules without eliminating incentives for innovation in the execution layer.
