Why Vitalik Buterin Says L2s Aren’t Scaling Ethereum Anymore

Ethereum co-founder Vitalik Buterin said recent developments mean the original conception of Layer 2 scaling within the ETH ecosystem is no longer viable. He said that the progress among many L2 networks has fallen short of earlier expectations, while the mainnet continues to scale directly. Slow Progress, Low Fees In a recent post on X, Buterin pointed to two important realities reshaping the debate. First, there is the slow and difficult progress of L2s toward “stage 2” decentralization and interoperability, and the fact that Ethereum’s mainnet has already achieved very low fees, with gas limits expected to rise significantly through 2026. Buterin reiterated that Ethereum scaling was originally defined as expanding block space that fully inherits Ethereum’s security. This means that all activity remains valid and censorship-resistant as long as the network operates. As such, systems that rely on multisig bridges or other forms of discretionary control cannot be considered extensions of Ethereum in this sense, even if they offer high throughput. The co-founder explained that this framing no longer holds because the blockchain no longer needs L2s to function as “branded shards,” while many L2s are either unable or unwilling to meet the security and governance requirements that such a role would imply. Buterin observed that some projects have explicitly stated they may never move beyond stage 1, not only due to technical concerns around zero-knowledge EVM safety, but also because regulatory or customer requirements necessitate ultimate control. While he said this may be appropriate for those projects’ use cases, it means they should not be described as scaling Ethereum under the original definition. Instead, Buterin suggested abandoning the idea that all Layer 2s should occupy the same category and be judged by the same criteria. He proposed that they be viewed as a broad spectrum of systems with varying degrees of connection to Ethereum. In this framing, some L2s...