I’ve been following Vitalik’s recent comments about Ethereum’s L2 strategy, and honestly, they’ve left me with mixed feelings as someone who’s spent the last few years building L2 infrastructure.
What Vitalik Actually Said
In February 2026, Vitalik stated that the “rollup-centric roadmap” from 2020 “no longer makes sense” in its current form. This is significant because the entire Ethereum ecosystem has been betting on L2s as the scaling solution for years now.
His main points:
- L1 is scaling faster than expected (gas limit going from 60M to 200M this year)
- Current L1 fees are below 2 gwei — incredibly cheap compared to 2021
- L2 decentralization progress has been “slower and more difficult than originally expected”
- Most L2s still run centralized sequencers despite being live for years
The Uncomfortable Questions
As an L2 engineer, I have to ask:
Did we bet on the wrong scaling strategy? We convinced developers to deploy on L2s, built complex bridging infrastructure, fragmented liquidity across 50+ chains… and now the base layer itself is becoming fast and cheap enough for most use cases.
What about all the L2 investment? Teams raised hundreds of millions to build L2 solutions. Developers spent years learning L2-specific tooling. Users bridged billions in assets. Was this all wasted effort?
Are centralized sequencers killing the L2 value proposition? If L2s are still centralized after 3+ years of operation, are they really offering anything beyond what a well-architected centralized database could provide?
What Vitalik Suggests Instead
He’s now proposing we view L2s as part of a “full spectrum” of networks with varying trust and security levels, rather than treating them as official Ethereum extensions. He wants L2s to develop distinct value propositions beyond simple scaling:
- Privacy-focused rollups
- Application-specific VMs
- Custom economics and governance
My Take
I think Vitalik is acknowledging what many L2 builders have quietly known for a while: the original thesis was too simple. “Just put everything on L2s” ignored the composability benefits of a unified L1 and underestimated how difficult sequencer decentralization would be.
But I don’t think L2s were a mistake. They’ve been an incredible experimentation layer for new VM designs, consensus mechanisms, and scaling techniques. The problem is we sold them primarily as scaling solutions when they should have been sold as innovation platforms.
Looking Forward
With L1 getting cheaper and faster, L2s need to justify their existence beyond “we’re 10x cheaper than mainnet.” That means:
- Actual decentralization (shared sequencers, based rollups)
- Real differentiation (privacy, compliance, app-specific features)
- Better interoperability (chain abstraction, unified liquidity)
What do you all think? Should Ethereum focus more on L1 scaling and let L2s be specialty chains? Or is the rollup-centric roadmap still the right call?
Sources: Yahoo Finance (Vitalik L2 comments), Cryptopolitan (L2 predictions 2026), Bitget News (gas limit increase)