Ethereum L2 Value Extraction

The economics problem

Layer-2 rollups were sold to the Ethereum community as a scaling solution that would preserve the base layer's value, with layer-2 activity eventually batch-settling on layer one, paying gas fees, and contributing to Ethereum's economic security. The observed reality by 2026 is closer to the reverse: one major layer-2 network was found to be paying more in licensing fees to the foundation that built its underlying technology stack than it was paying in Ethereum layer-one fees, by roughly two to three orders of magnitude.¹ The fee flow routes significant revenue to the technology licensor and keeps most transaction fees for the layer-2 network itself, while the base layer receives a comparatively tiny share through blob fees. The layer-2 network rides the base layer's brand, security model, and network effects without proportionally compensating it, a dynamic CZ reads as one reason Vitalik Buterin faced real pressure over the base layer's price performance and eventually adjusted course.

The centralization problem

Layer-2 networks were supposed to scale without sacrificing decentralization, but nearly all major ones operate with a single, centralized sequencer, one entity that orders transactions, sets fees, and holds unilateral power over the rollup's operation. The formal milestone representing genuinely decentralized operation in the layer-2 maturity framework has not been reached by any major network as of 2026, even as the base-layer community continues to push for full decentralization elsewhere.¹

The unilateral exit problem

Defenders of the layer-2 model point to unilateral exit, the ability to replay a layer-2 transaction directly on the base layer by bypassing the sequencer, as the underlying safety guarantee. For simple asset transfers this works in principle. For active financial operations such as collateral top-ups or defending a position against liquidation, timing is everything, and if a sequencer censors a user for even a minute, a unilateral exit to the base layer arrives too late to matter. That gap is precisely why some builders of neutral financial infrastructure have chosen not to run on a layer-2 network at all, judging it insufficiently decentralized to serve as credibly neutral trading infrastructure.¹

The base-layer response

Vitalik Buterin's public signal of renewed base-layer focus in 2025 and 2026 is read as a direct response to both the economics problem and the centralization problem: the layer-2 roadmap was not delivering what had been promised for either.¹ One concrete piece of that renewed focus is censorship resistance built into the base layer itself: inclusion-list mechanisms force block proposers to include eligible transactions, directly answering the sequencer-censorship timing risk described above, though not for free. A formal result shows any censorship-resistant consensus protocol in this design space needs at least two additional communication rounds beyond the standard three-round baseline once a meaningful share of validators may be faulty, a mathematical lower bound rather than an engineering shortcut. The reframing offered is that those two extra rounds do not slow down the uncensored path, they instead cap censored-transaction latency at a fixed bound instead of leaving it open-ended for however long censorship lasts, so censorship resistance actually reduces the latency a user experiences when they need censorship-resistant inclusion.²

A complementary supply-side argument names the layer-2 problem directly and proposes shipping the base layer itself faster rather than routing everything through rollups: native rollups, layer-2 networks that inherit the base layer's security and censorship resistance rather than relying on a single trusted sequencer, are named as a near-term roadmap item, aimed squarely at both the value-extraction and centralization problems described above. The broader argument for shipping the base layer faster is that speed helps regardless of one's position in the layer-1-versus-layer-2 debate, and that many apparent either-or tradeoffs dissolve into both once the base layer reaches closer to the efficient frontier of what is technically possible.³

Why it matters

The broader implication is that the layer-2 era may turn out to be a detour rather than the final architecture, with the base layer remaining where fundamental value accrual actually happens, and layer-2 networks ending up as institutional tools, such as banks running their own instance of a rollup stack or national chains, rather than the primary consumer experience layer.¹ Part of the market's migration away from Ethereum layer-2 networks toward alternative chains for decentralized finance activity is explained by exactly the centralization and economics failures described here.