In recent weeks, the biggest conversation in the Ethereum community has been Vitalik Buterin’s public rethink of the scaling roadmap.
Vitalik’s stance was unusually direct: as Ethereum mainnet (L1) improves its own scaling, the five-year-old roadmap, which positioned L2s as the primary scaling method, has now become obsolete.
Some readers took this as pessimism—or even a rejection—of L2s. But read in context of L1 upgrades, decentralization frameworks, and the latest technical discussions around Native/Based Rollups, Vitalik’s stance is less a rejection of L2s and more of a 'course correction' aimed at restoring Ethereum's original strategic focus.
Ethereum isn’t abandoning L2s—it’s clarifying the split. L1 focuses on being the most trusted settlement layer; L2s specialize and differentiate, so the ecosystem’s long-term strategy stays anchored to mainnet.
1. Have L2s fulfilled their historical role?
In the cycle before last, L2s were widely seen as Ethereum’s lifeline.
In the early Rollup-Centric roadmap, the split was clear: L1 provided security and data availability; L2 delivered scale and low fees. When gas hit tens of dollars, it was one of the few workable paths.
But the way things unfolded was more complicated.
L2BEAT’s latest data shows that L2s (broadly defined) now number over 100. But growth in count hasn’t meant maturity—and most still move slowly on decentralization.
A key detail: back in 2022, Vitalik criticized the “training wheels” model—Rollups that rely on centralized ops and human backstops. L2BEAT tracks this via a prominent metric on its homepage: Stage.
The framework groups Rollups into three decentralization stages—Stage 0, Stage 1, and Stage 2—reflecting how much they still depend on “training wheels” and manual intervention.
Vitalik also noted that some L2s—driven by regulation or business needs—may stay at Stage 1 indefinitely, relying on a security council for upgrade control. In that case, they resemble “secondary L1s” with a bridging layer, not the “branded shards” once imagined.
If sequencing, upgrade rights, and final adjudication remain concentrated within a few entities, it not only contradicts Ethereum’s vision of decentralization but also risks turning these L2s into parasitic entities that extract value from the mainnet without contributing to its core security.
Meanwhile, L2 proliferation has intensified a structural issue users have felt for years: liquidity fragmentation.
Activity that once concentrated on Ethereum is now split across separate ecosystems, creating isolated pools of liquidity—and more chains generally means more fragmentation, which wasn’t the original goal of scaling.
That’s why Vitalik emphasizes the next step isn’t more chains, but deeper integration: using more standardized designs and protocol-native security to reinforce L1 as the most trusted settlement layer.
In this framing, scaling isn’t the only goal. Security, neutrality, and predictability become core strengths again. L2s matter less by how many exist, and more by how tightly they integrate with mainnet—and how well they innovate in focused niches (privacy VMs, ultra-high throughput, or specialized execution for non-financial apps like AI agents).
This matches comments from Ethereum Foundation Co-Executive Director Hsiao-Wei Wang at Consensus 2026: L1 should host the most critical activity as the safest settlement layer, while L2s specialize to deliver the best possible user experience.
2. Native Rollups: Based Rollup + preconfirmations as the path forward?
In this wave of rethinking, Based Rollups could take center stage in 2026.
If the keyword of the last five years was “Rollup-Centric,” the discussion is now narrowing to a more specific question: can a rollup be 'enshrined' directly within the Ethereum protocol, rather than operating as a detached external add-on?
That’s why “Native Rollup” is getting attention. If Native Rollups are the end state, then Based Rollups are the most practical step we have today.
The key difference is sequencing. Based Rollups don’t rely on an independent (often centralized) sequencer. Instead, transaction ordering is driven by Ethereum’s L1 block production—bringing rollup execution closer to the protocol’s security guarantees, rather than running as a separate system.
For users, the rollup feels embedded in Ethereum. It inherits L1’s censorship resistance and liveness—and, crucially, improves synchronous composability: within a Based Rollup block, you can directly tap into L1 liquidity and achieve seamless, atomic cross-layer transactions.
But there’s a trade-off. If you fully follow L1’s 12-second slots, interactions can feel sluggish. And even after inclusion, finality can take roughly two epochs (~13 minutes), which is too slow for many financial use cases.
In the same post, Vitalik pointed to a January proposal that argues for a hybrid: keep low-latency sequenced blocks, produce a based block at the end of each slot, submit it to L1, and combine this with preconfirmations to reach synchronous composability.
Here, preconfirmations mean: before a transaction reaches full L1 confirmation, a designated party (for example, the L1 proposer) provides a credible commitment that it will be included. This maps closely to Ethereum Interop’s Project #4: Fast L1 Confirmation Rule.
The goal is to provide applications and cross-chain protocols with a 'strong and verifiable' L1 confirmation signal within 15–30 seconds, bypassing the 13-minute wait for full finality.
Mechanically, this doesn’t add a new consensus process. It reuses PoS attestations that already happen each slot: once a block gathers enough widely distributed votes early, it can be treated as very unlikely to be rolled back under reasonable attack assumptions—even before finality.
In short, it doesn’t replace finality. It adds a protocol-recognized “strong confirmation” before finality—so bridges, intent solvers, and wallets can move forward within 15–30 seconds based on a safer, protocol-level signal.
This layered model creates clear trust tiers between security and perceived speed—opening the door to a more seamless interoperability experience.(Further reading: “Ethereum’s "Second-Level" Evolution: How Interop Eliminates Wait Time”)
3. What’s next for Ethereum?
From a 2026 vantage point, Ethereum’s focus is shifting—from maximizing throughput to prioritizing unity, layered design, and protocol-native security.
Recently, several L2 leaders said they’re open to a Native Rollup path to improve network-wide consistency and coordination. The signal is clear: the ecosystem is moving away from “more chains” and back toward “a more unified protocol.
As L1 strengthens and Based Rollups plus preconfirmations become more real, performance stops being the only constraint. A more practical bottleneck stands out: wallet UX and onboarding friction.
This echoes what imToken highlighted in 2025: as infrastructure fades into the background, mass adoption hinges on the entry-point experience.
Zooming out, Ethereum’s next phase may center on three structural shifts:
- Account abstraction and lower onboarding friction: Ethereum is pushing Native Account Abstraction (Native AA). Smart contract wallets may become the default, replacing seed phrases and EOA-style addresses—so for users of wallets like imToken, getting started could feel as easy as signing up for a social app. (Further reading: “From EOA to AA: Will Web3’s Next Leap Happen at the Account Layer?”)
- Privacy and ZK-EVM: Privacy is no longer a niche feature. As ZK-EVM matures, Ethereum can stay transparent while providing the on-chain privacy commercial use cases need—an important edge as public chains compete. (Further reading: “ZK Dawn: Is Ethereum’s Endgame Accelerating?”)
- On-chain sovereignty for AI agents: By 2026, transactions may be initiated by AI agents, not humans. The challenge is trustless standards—proving agents execute user intent rather than being steered by third parties. Ethereum’s decentralized settlement layer could serve as the rules arbiter for an AI economy. (Further reading: “A Passport to the AI Agent Era: Why Ethereum is Betting Big on ERC-8004”)
So, did Vitalik “reject” L2s? Not really. He’s rejecting an inflated, mainnet-detached fragmentation narrative—and treating that correction as a new beginning.
Moving from the big dream of “branded sharding” to the more concrete engineering of Based Rollups plus preconfirmations could further strengthen L1 as Ethereum’s trust anchor.
But it also raises the bar: the L2s that last will be the ones aligned with Ethereum’s next-stage principles—and that grow with mainnet, not away from it.