Ethereum’s Plummeting Fees Are Forcing a Rethink of Layer‑2’s Purpose

Ethereum’s base layer has quietly rewritten the assumptions that justified the Layer‑2 gold rush. With mainnet fees hovering around $0.35 on average and gas prices frequently in fractions of a gwei, the original idea of rollups as near-identical, cheap “shards” of Ethereum no longer fits reality. Vitalik Buterin’s recent research posts crystallize that shift: most existing Layer‑2 (L2) chains now need a new, clearer job description.

For crypto investors and Ethereum builders, this isn’t a story of L2s being abandoned. It’s a story of re‑tiering: a sharper separation between rollups that truly extend Ethereum’s security, those that intentionally retain centralized control, and those that exist to specialize rather than simply undercut on fees.

The Rollup Roadmap Meets a Cheaper L1

Back in October 2020, Ethereum’s rollup‑centric roadmap emerged from a simple, urgent premise: mainnet was too expensive, and it couldn’t scale for most users. Gas prices were spiking, some applications were effectively priced out, and the ecosystem committed to going “all‑in on rollups” in the near and medium term. L2s were framed as branded shards of Ethereum, all broadly sharing the same security story and competing largely on cost and user experience.

Today, that premise is much weaker. On-chain data shows the opposite of the 2020 crisis that motivated the roadmap. The seven‑day average Ethereum transaction fee sits around $0.35, and gas prices at times fall into the fractions of a gwei. On Jan. 16, 2026, Ethereum even set an all‑time high of 2,885,524 transactions in a single day, pairing record activity with low fees—a “busier and cheaper” regime instead of “congested and expensive.”

At the protocol level, Ethereum’s execution capacity is also rising. In late 2025, validators signaled support to raise the block gas limit to roughly 60 million, up from the long‑standing 30 million level. With 12‑second blocks, that puts throughput around 5 million gas per second. Community conversations have floated even higher aspirational targets—up to 180 million gas—though this remains directional, not a committed roadmap.

For L2s, this matters directly. The original justification that “L1 can’t scale for most users” is now materially weaker. Layer‑2 doesn’t need to be a uniform set of low‑cost clones; it can be a spectrum of different trade‑offs in security, sovereignty, and specialization layered on top of a cheaper and gradually scaling base chain.

Decentralization Stalled: What L2BEAT’s Stages Reveal

To understand how far today’s L2s are from the idealized rollup vision, investors and builders increasingly rely on L2BEAT’s Stages framework. It offers a concrete way to classify rollups based on how much decentralization and trust minimization they actually deliver.

Stage 0 represents rollups with “training wheels” firmly attached. These systems still have significant trust assumptions—upgrade keys, privileged operators, or emergency paths that can override proofs and user guarantees.

Stage 1 marks partial decentralization. Rollups at this stage offer stronger escape hatches and proof guarantees, but users still must trust upgrade mechanisms, governance structures, or security councils to some degree. There are meaningful discretionary powers in the system.

Stage 2 is the endgame that many early narratives implicitly promised: “no training wheels.” Here, critical safety properties are enforced by code rather than by people. Upgrade powers are heavily constrained and delayed, proof systems cannot be trivially overridden, and users rely primarily on cryptography and protocol rules—not discretionary actors—for security.

The actual distribution of value across these stages is stark. According to L2BEAT’s rollup scaling summary, roughly 91.5% of tracked value sits in Stage 1 rollups, 8.5% in Stage 0, and about 0.01% in Stage 2. The top three rollups by value account for around 71% of the total, meaning the practical progress toward Stage 2 security is overwhelmingly dependent on a handful of major projects, not on the many smaller experimental chains.

The main blockers are straightforward but hard: can proof systems be overridden, and how constrained are upgrades? In the largest rollups, upgrade discretion and emergency powers remain common. Security councils, foundations, or multisigs can often push changes with little or no delay, and some have explicit emergency paths that bypass normal exit windows. Moving away from these arrangements has proven much slower and more complex than early 2020–2021 optimism suggested.

Some teams have openly said they may not want to go beyond Stage 1 at all, citing not only technical risks—such as concerns around zkEVM safety—but also regulatory requirements that demand tight operational control. As a product stance, that is consistent: for certain compliance‑focused or institutional users, controllability is a feature. But it also clarifies that many of these chains are not “scaling Ethereum” in the original sense—they are more akin to controlled execution environments that happen to post data to Ethereum.

From “Branded Shards” to a Spectrum of Layer‑2 Designs

Against this backdrop, Vitalik Buterin’s reframing of L2s is less about repudiating the rollup roadmap and more about aligning it with reality. Instead of treating all L2s as near‑identical shards, his recent Ethereum Research posts describe a spectrum.

At one end are rollups that try to inherit Ethereum’s security as fully as possible. These are chains that aim for Stage 2: code‑enforced safety, minimal governance discretion, and strong guarantees for ETH and Ethereum‑issued assets. They might not just be EVM clones—some could be privacy‑preserving systems or non‑EVM environments—but their defining trait is their deep, trust‑minimized anchoring to Ethereum.

At the other end of the spectrum are chains with weaker or more selective connectivity to Ethereum. They may retain strong upgrade keys, embed regulatory controls, or act more like app‑specific environments. For these, Buterin’s minimum bar is explicit: if you handle ETH or Ethereum‑issued assets, you should at least meet Stage 1. If you don’t, you’re functionally a separate L1 with a bridge and should call yourself that.

The harder bar is differentiation. In a world where mainnet is cheaper and “generic cheap EVM” is commoditized, L2s need to be best‑in‑class at something specific: privacy, application‑specialized efficiency, extreme scaling beyond even an expanded L1, or radically different designs for non‑financial applications such as social networks or identity systems. Built‑in oracles, decentralized dispute resolution, and ultra‑low‑latency sequencing for trading are cited as examples of features that could justify an L2’s existence without pretending to be just another shard.

Technically, one potential path to support this spectrum is a “native rollup precompile” inside Ethereum. This would let the base protocol verify a standard zkEVM proof directly, so rollups that are “EVM plus extras” could have their core EVM behavior verified trustlessly at L1 while proving their custom extensions separately. Research posts published on Jan. 16 and Feb. 2 explore how such primitives could enable synchronous composability—where contracts on different rollups interact within a single transaction.

For now, though, these ideas are research directions, not shipped protocol features. Investors should treat them as signals of where Ethereum might go, not as guarantees of near‑term upgrades.

Three Strategic Buckets for the Next Wave of Rollups

If this reframing takes hold, the sprawling L2 landscape is likely to sort into three broad strategic buckets.

1. Stage‑2‑chasing settlement rollups. These projects interpret their mandate as “scale Ethereum” in the strictest sense. Their roadmap centers on maximizing security inheritance from mainnet—pursuing code‑enforced guarantees, tightly constrained governance, and robust escape hatches. For users and DeFi protocols, these are the chains that aim to feel as close as possible to Ethereum itself, but with more throughput and potentially distinct execution models.

2. Regulated or controlled execution environments. This category is for L2s that optimize for compliance, permissioning, and institutional requirements. By design, they may never progress beyond Stage 1 because they rely on strong administrative controls for risk management, regulatory reporting, or operational responsiveness. Their honest value proposition is not “pure decentralization” but “Ethereum‑adjacent compute with explicit governance and policy controls.”

3. Specialized chains for latency, privacy, or non‑financial use cases. These L2s justify themselves by focus, not by generality. Privacy rollups that use zkProofs to hide transaction details, ultra‑low‑latency sequencers for trading, app‑specific chains with tailored throughput for gaming or AI, and social or identity systems with very different state models all fall into this bucket. They may or may not be EVM‑compatible, and they do not need to be fully financial to matter; they just need to deliver something users cannot get elsewhere.

Major projects like Arbitrum One, Optimism’s OP Mainnet, Base, zkSync Era, and Starknet will each have to decide where on this spectrum they want to land. The ecosystem is big enough to support all three buckets, but the idea that “every L2 does the same job” is already fading. For investors, that means less thinking in terms of “L2 market share” and more in terms of which specific category—and guarantee set—a given rollup is targeting.

Implications for Users, Wallets, and Application Teams

For end users, the change is subtle but important: trust assumptions can no longer be taken for granted. Escape hatches, fraud or validity proofs, upgrade delays, emergency powers, and censorship resistance are becoming product differentiators, not baseline features. Two rollups can both brand themselves as “Ethereum L2s” while offering very different failure modes in a crisis.

This raises the bar for wallets and interfaces. To remain credible, they will need to surface L2BEAT‑style staging information and clearly label the trust assumptions behind each network. Instead of a simple list of “supported chains,” users may see indicators for training wheels, security councils, exit guarantees, and proof verification behavior.

For builders deploying applications, the takeaway is that “cheap EVM” no longer constitutes a moat. With L1 fees down and standard rollup infrastructure broadly available, differentiation moves elsewhere: advanced privacy, custom virtual machines, ultra‑low‑latency sequencing for specific markets, app‑specific throughput tuning, or explicit compliance and permissioning for institutional users. Chains that lean on Ethereum for credibility while retaining extensive discretionary control will increasingly be judged on whether they describe themselves accurately.

Market narratives will evolve as well. Rival L1 communities have long questioned whether L2s “really” inherit Ethereum’s security. The fact that most value still sits on Stage 1 systems with strong governance powers makes that critique more pointed. The Ethereum ecosystem’s own public acknowledgment of this reality—through L2BEAT’s data and Buterin’s reframing—suggests future debates will focus more on actual guarantees than on aspirational marketing.

Rethinking “Scaling Ethereum” in a Multi‑Tier Future

Despite the dramatic language that “most L2s lack purpose,” what is emerging is less a collapse of the rollup thesis and more a stratification of it. Ethereum is unlikely to see a single, uniform L2 revolution. Instead, it is experiencing a re‑tiering of the entire stack.

The old picture assumed: expensive, capacity‑constrained L1; a relatively homogeneous field of L2 “shards” that all looked similar; and competition primarily on fees and UX. The new picture is different. L1 is cheaper and expanding in capacity, while L2s are diverging in security models, governance, and use cases faster than they are converging on a single standard.

Under this reality, the minimum bar for any chain that touches ETH or Ethereum‑issued assets is clear: reach at least Stage 1 and be transparent about your governance and upgrade paths. Beyond that, the differentiation bar is also clear: be the best at something concrete, whether that’s trust‑minimized scaling, controlled environments for institutions, or deeply specialized functionality that mainnet and generic rollups can’t match.

Research into native verification primitives and synchronous composability shows where Ethereum might go to make cross‑rollup interactions safer and more fluid. But these are trajectories, not live features, and investors should treat them accordingly. For now, the practical opportunity lies in understanding which L2s align with which guarantees—and whether those guarantees match the needs of the users, applications, or capital they aim to attract.

Ethereum’s rollup‑centric roadmap hasn’t been discarded; it has been upgraded to reflect a cheaper, more capable L1 and a more diverse L2 ecosystem. The job description for a Layer‑2 has changed. Chains that can clearly articulate their role on this new spectrum—and back it with concrete security properties—are the ones most likely to retain relevance as the market recalibrates.