The end of “cheap ether”: why Buterin is rewriting L2’s future—and who may be left behind

Why “cheap ether” on L2 is ending—and what it means for rollups.

For years a stereotype ruled crypto: expensive Ethereum is for whales; ordinary users belong on the “second floor”. Billions of dollars fuelled ecosystems whose business model was simple intermediation: projects bought mainnet blockspace wholesale to offer clients cheaper transactions.

The Rollup-centric strategy looked unassailable until February 2026, when Ethereum co-founder Vitalik Buterin signalled a shift in direction: the base layer is gearing up for a technological leap, and the network no longer needs centralised “branded shards”.

With the new Strawmap scaling plan, the market stands on the brink of sweeping change. Here is why the idea that only the second layer offers cheap transactions is obsolete—and how leading L2s may respond.

A turn in direction and the scaling paradox

Since 2020 Ethereum’s roadmap was built around rollups. The assumption was that the mainnet would remain an expensive, slow but secure foundation, while most user activity would shift to blockchains such as Optimism, Arbitrum and Base.

In early 2026 Buterin effectively ripped up that social contract. His policy pronouncements on X were a cue to radically reassess architectural priorities.

There have recently been some discussions on the ongoing role of L2s in the Ethereum ecosystem, especially in the face of two facts:

* L2s’ progress to stage 2 (and, secondarily, on interop) has been far slower and more difficult than originally expected
* L1 itself is scaling,…

— vitalik.eth (@VitalikButerin) February 3, 2026

For investors this is a reason to reassess the market capitalisation of L2 tokens whose utility rested on the narrative of “saving” Ethereum from high fees. For developers it reads like an ultimatum: offer genuinely distinctive products or end up as a redundant link in the gas-resale chain.

A paradox has emerged: while the industry spun up hundreds of rollups to relieve the mainnet, Ethereum itself no longer needs them in their old guise. Behind the shift are L2s’ structural decentralisation problems—and the base chain’s readiness for aggressive scaling.

A crisis of trust and the “training wheels”

The trigger for Buterin’s discontent was technical stagnation and L2 teams’ reluctance to embrace real decentralisation. Back in 2022 he proposed a three-step rollup-maturity scale: from Stage 0 (full developer control) to Stage 2 (security ensured solely by code and cryptography).

By early 2026 it became clear that progress towards Stage 2 was far slower than the Ethereum Foundation had hoped. Most major networks are stuck at the first stage. They keep the “training wheels”—security councils, multisigs and admin keys that allow manual changes to blockchain state.

The issue is not just technical difficulty. Buterin pointed to a structural conflict of interest: many projects are deliberately slowing the handover of control to their communities.

Motivations range from preserving business models to meeting regulatory demands. Some developers openly refuse sweeping changes to suit institutional clients who want the option to censor transactions or freeze assets. That sits uneasily with Ethereum’s principles.

Today’s L2s face several core problems:

• an illusion of continuity: projects trade on Ethereum’s brand while remaining isolated systems with “optimistic” bridges, week-long withdrawal delays and centralised sequencers;
• technological stagnation and copy-paste: instead of new architectures the market is awash with EVM clones that bring no advances in performance or privacy;
• dependence on multisigs: control over smart contracts remains in the hands of a few. Security depends on people, not mathematics;
• poor interoperability: liquidity is fragmented across hundreds of rollups, degrading the user experience.

In Buterin’s view, if a project cannot—or will not—reach Stage 2, it should not be considered a full-fledged part of Ethereum.

Ambitious plans

While second-layer projects have bogged down in partial centralisation, Ethereum Foundation researcher Justin Drake unveiled Strawmap—a sweeping document sketching the network’s new horizons and tempo.

The roadmap outlines roughly seven planned hard forks through 2029, with updates every six months. Drake says intensive use of artificial intelligence during development could shorten those timelines.

Consensus-layer upgrades follow a “stellar” naming scheme (Altair, Bellatrix, Capella, Deneb, Electra, Fulu). Only two have confirmed names—Glamsterdam and Hegota—and both are slated for rollout in 2026.

Stra wmap effectively turns Ethereum into a direct competitor to its own rollups. The stated performance targets look menacing for projects built on reselling cheap gas:

• L1 throughput up to 10,000 TPS: achieved via a limit of 1 GB of gas per second, thanks to zkEVM at the protocol level and real-time generation of cryptographic proofs;
• L2 throughput up to 10m TPS: reached using data-availability sampling that lets second-layer networks publish up to 1 GB of data per second;
• native rollup precompile: a protocol-integrated function for verifying ZK proofs. It lets L2s upgrade in lockstep with the mainnet and fix critical bugs via a mainnet hard fork.

Buterin placed special emphasis on reducing block times. Twelve seconds is too slow for modern decentralised applications.

Speed-ups are planned in stages: first from 12 to 8, then to 6, 4 and 2 seconds. The key condition is optimising P2P communication between nodes so blocks propagate without overloading the network. In parallel, transaction finality is to be cut from 16 minutes to 6–16 seconds.

That will require substantial execution-layer changes, including a move to post-quantum cryptography using hash-based signatures.

The great split: L2 developers strike back

Claims that the current L2 model has lost its relevance sparked heated debate among project leaders. The focus quickly shifted from technicalities to defending market capitalisations.

Stephen Goldfeder, co-founder of Offchain Labs, took the hardest line. He stressed that Arbitrum was never conceived as “a service for Ethereum”. The mainnet is merely a settlement security layer; L2 networks are independent economies.

As evidence, Goldfeder recalled peak-load periods when Arbitrum and Base together processed over 1,000 TPS while the mainnet lingered at 15–20 TPS. He warned that if Ethereum devalues rollups’ role, institutions will launch sovereign L1s—ushering in full-blown liquidity fragmentation.

Base creator Jesse Pollak sought conciliation. He agreed that second-layer solutions should not remain “just cheap ether” and shifted the focus to user experience. Teams are betting on account abstraction, simpler onboarding and distinctive features. Even with a fast base layer, users need smooth interfaces the mainnet, by design, cannot provide.

StarkWare co-founder Eli Ben-Sasson responded wryly. He noted that Buterin’s “new vision” describes what Starknet has been building for years—a system based on ZK-STARK proofs, high performance and a path to decentralisation (Stage 2).

Beneath the sparring lies a fundamental problem: L2 teams now have to re-justify the value of their tokens. If scaling is executed successfully at the base layer, many projects risk fading from view.

Buterin’s scepticism toward the “rollup-centric” model jars with marketing strategies that for years leaned on symbolic proximity to the Ethereum mainnet as the chief source of legitimacy.

Spectrum, not standard: a new taxonomy for the ecosystem

Buterin suggests abandoning a rigid L1-versus-L2 dichotomy in favour of a “spectrum of possibilities”. Projects will be classified by technical architecture and real security contribution, not advertising claims.

He highlights three directions:

• app-specific systems: solutions with unique logic, such as game engines or decentralised identity. Ethereum serves as the base layer, but these systems do not seek full EVM compatibility;
• institutional networks: corporate blockchains that may retain partial centralisation for regulatory reasons, but must publish ZK proofs or state data on the mainnet for transparency;
• non-financial applications: social platforms, AI agents and reputation systems. Developers are urged to build products resilient to corporate or state pressure.

The foundation for all applications should be the CROPS concept, proposed in early March 2026:

• censorship resistance;
• open source;
• privacy;
• security.

Privacy is singled out as the bedrock of freedom. The Ethereum Foundation is bolstering teams integrating privacy tools into the base layer. One such effort is Kohaku—a modular open-source stack for building secure wallets.

Buterin sees the future of interfaces in deep integration with AI tools. Neural networks will simulate transactions before signing: users see a clear outcome and confirm manually only in critical cases.

New advances and the “adoption paradox”

The Ethrex team, together with the Ethereum Foundation and L2BEAT, has already released code and documentation for a native rollup execution environment.

The prototype implements EIP-8079 on the Ethrex client and introduces a new mechanism—the precompile EXECUTE. The function lets L2 blocks be re-executed directly on the mainnet, so the mainnet itself attests to their correctness.

At present all L2s in the Ethereum ecosystem validate via fraud proofs or ZK schemes. Native rollups propose an alternative: the base chain recomputes network state through EXECUTE. External proofs are no longer required. If the concept proves out, rollups could inherit security parameters and software upgrades directly from Ethereum.

Despite developer activity, the cryptocurrency’s price continues to stagnate. CryptoQuant analysts noted a gap between on-chain demand and the price of ETH, which has fallen by more than 50% from recent peaks.

The divergence is not limited to user numbers. CryptoQuant recorded a sharp surge in smart-contract and automated-protocol activity.

Experts attribute the gap to capital outflows—price action is driven by flows, not user growth. Owing to this “adoption paradox”, Ethereum risks sliding to $1,500, said researcher Julio Moreno. Absent a decisive market shift, the asset could reach that level in late Q3 or early Q4.

Even so, record on-chain activity may be driven by mass “address poisoning”. Etherscan data show that after the Fusaka upgrade, the number of dust USDT transfers jumped by 612%. One reason for the rise in scams is lower network fees.

Back to basics—or forced evolution?

The takeaway: the era of earning by simply cloning EVM infrastructure is ending. A new vector for the ecosystem is at hand.

The narrative of “saviour rollups” is losing relevance. The network is evolving either into a global computer with 10,000 TPS on the base layer, or into a verification substrate for thousands of specialised platforms adhering to CROPS.

Buterin’s metaphor about “taking off jackets and ties” is a call to return to cypherpunk ideals. The co-founder expects more flexibility and less corporate control. Nodding to the internet culture around the Milady NFT collection, he notes that to preserve freedom one sometimes has to leave the comfort zone and symbolically “spill wine on your chest”, shedding old constraints.