Ethereum Layer 2 decentralization falls short of expectations: What does Vitalik's statement imply?

In May 2026, the Ethereum Foundation concluded a week-long Soldøgn Interop technical sprint on the Svalbard Archipelago in Norway, with over a hundred core developers officially wrapping up collaborative efforts on hardening the Glamsterdam upgrade. This meeting not only achieved the core technical goals of Glamsterdam but also simultaneously confirmed a pivotal shift in the direction of the Hegotá upgrade — from the original scalability roadmap to a “cleanup and hardening” fork focused on addressing technical debt through “culling and strengthening.” However, at nearly the same time, another widely circulated community assessment warrants deeper reflection: Vitalik Buterin explicitly acknowledged that, within Ethereum’s Rollup-centric roadmap, the progress of Layer 2 decentralization is “far slower than expected.” This reality, intertwined with Ethereum’s own rapid layer 1 (L1) scalability expansion, is reshaping the fundamental logic of Ethereum’s overall scalability pathway.

Why Ethereum Layer 2 Decentralization Progress Continues to Fall Short of Expectations

In February 2026, Vitalik publicly stated that the roadmap established five years prior, which positioned Layer 2 as Ethereum’s primary scalability solution, is no longer applicable. His core judgment rests on two facts: first, that the “progress of Layer 2 towards higher decentralization stages is far slower and more difficult than anticipated”; second, that Ethereum’s L1 scalability has significantly outpaced initial projections.

From a decentralization tiered framework, L2BEAT categorizes Rollups into three stages—Stage 0 (fully centralized), Stage 1 (limited reliance on multi-signature governance), and Stage 2 (fully decentralized, relying solely on code and cryptography). As of early 2026, the vast majority of top-tier Layer 2s remain at Stage 0 or Stage 1, not yet achieving full decentralization. Even the few that have advanced to Stage 1 are still far from meeting the “no backdoor control” standard required for Stage 2.

This slower-than-expected progress is driven by both technical and economic factors. Some L2 teams explicitly state that, constrained by regulatory requirements or business models, they may never pursue full decentralization. The core commercial model for L2 operators—revenue from sequencers—means that decentralizing sequencers to reduce reliance on centralized entities entails ceding economic incentives, which in turn limits the pace of decentralization in practice.

How Sequencer Centralization and Multi-Signature Bridges Expose Three Structural Bottlenecks

Dissecting the structural causes behind the lagging decentralization of L2, three interconnected issues emerge as key bottlenecks.

First is the centralization of sequencers. Currently, most mainstream L2s depend on a single centralized sequencer to package and order transactions. While this approach offers high efficiency and low costs, it introduces vulnerabilities such as weak censorship resistance and single points of failure. The sequencer controls transaction ordering, enabling it to extract maximum MEV (Miner Extractable Value) and potentially conduct transaction censorship—directly contradicting Ethereum’s core principle of decentralization.

Second is the lag in implementing fraud proofs and validity proofs. Optimistic Rollups rely on challenge windows (typically 7 days), requiring users to trust the L2 operator for extended periods. While ZK Rollups theoretically offer instant finality, generating validity proofs involves highly specialized circuits and complex verification processes. Moreover, each Ethereum hard fork that changes EVM behavior necessitates synchronized upgrades across all L2s’ proof systems, incurring substantial overhead.

Third is the fragmentation of cross-chain liquidity. By early 2026, over 50 major Rollup networks with total value locked (TVL) exceeding $45 billion have emerged, but funds and users are dispersed across multiple chains and bridges, leading to increasing liquidity fragmentation. Most L2s connect to Ethereum L1 via multi-signature bridges—cross-chain asset transfer mechanisms controlled by multi-signature contracts. Vitalik has criticized this approach: a chain capable of 10,000 TPS with a connection relying on a single multi-signature bridge does not truly extend Ethereum’s security but merely creates an independent trust-based platform. The widespread use of multi-signature bridges indicates that most Rollups do not inherit Ethereum’s security guarantees but depend on centralized control to operate.

The Launch of Glamsterdam Devnet and How ePBS Addresses Scalability and Security Challenges

The launch of Glamsterdam devnet marks one of the most significant milestones in Ethereum’s 2026 roadmap. Before Soldøgn Interop concluded in early May, glamsterdam-devnet-2 achieved stable operation, with multi-client support for ePBS (protocol-embedded Proposer-Builder Separation) completing end-to-end cross-client testing, covering “almost all client implementations.”

ePBS’s core value lies in separating block proposal rights from block construction rights, embedding a standardized MEV supply chain mechanism at the protocol level. Previously, block building depended on external relays, which posed centralization risks; ePBS integrates block construction and validation into the protocol rules, greatly reducing MEV manipulation potential. It also restructured the slot architecture, adding clear deadlines for block construction and proposal, providing a larger buffer for subsequent security enhancements such as increasing gas limits.

Glamsterdam has also set a target gas limit of 200 million units post-upgrade. Combined with the time-structure optimizations of ePBS and the parallel verification enabled by block-level access lists (BAL), developers now have a more concrete engineering baseline for the scalability of the mainnet in 2026.

Fusaka Scalability Milestone Achieved and Structural Breakthroughs in Data Availability

The Fusaka upgrade was officially activated on December 3, 2025. Its core innovation, PeerDAS (EIP-7594), embeds data availability sampling capabilities into the protocol layer. By allowing nodes to store only a subset of Blob data rather than the full dataset, PeerDAS theoretically increases Blob capacity by approximately 8 times, providing Layer 2 networks with more ample data availability space. This change directly reduces hardware requirements for running nodes—maximally decreasing Blob bandwidth needs for typical operators by up to 80%.

Another key significance of Fusaka is establishing Ethereum’s development rhythm of “two hard forks per year.” From the Pectra upgrade in May 2025 to Fusaka in December 2025, just seven months apart, this marks a shift from lengthy, controlled upgrade cycles to accelerated iteration.

However, Fusaka remains primarily focused on scalability; core features related to decentralization and censorship resistance—such as further upgrades—have been postponed to subsequent upgrades. Strategically, this means prioritizing scalability first, with governance and decentralization improvements coming later—a sequence that continues to spark ongoing community debate.

Why Hegotá’s Shift Toward “Cleanup and Hardening” Instead of Further Scalability

Hegotá is positioned as a major second-phase upgrade in the second half of 2026, but its focus has shifted significantly—from the original “Scalability Roadmap” to a “cleanup and hardening” fork. Features like Fork-choice Inclusion Lists (FOCIL), account abstraction (AA), and alternative signature schemes have been incorporated into Hegotá’s scope.

This shift is driven by the realization that, after Fusaka’s data availability expansion and Glamsterdam’s throughput improvements, Ethereum’s L1 scalability has far exceeded the baseline set in the 2020 Rollup-centric roadmap. Vitalik notes that the low transaction fees and the continuous increase in gas limits on L1 mean that “the layer 1 scalability speed is far beyond expectations.” In this context, the value proposition of Layer 2 is being recalibrated—no longer as Ethereum’s “official sharding,” but as a layer that must offer capabilities that L1 cannot provide, such as privacy, ultra-low latency, or application-specific optimizations, to justify its existence.

The inclusion of FOCIL, a key feature for enhancing censorship resistance, into Hegotá signifies giving core developers more time to refine protocol-level mandatory transaction inclusion mechanisms. This is an infrastructural task that users cannot directly perceive but is crucial for protocol fairness.

Can Based Rollup and Pre-Confirmation Mechanisms Break the Deadlock?

To address issues of sequencer centralization and cross-chain interoperability, Based Rollup offers an alternative approach: the block ordering rights are assigned to Ethereum L1 validators rather than independent sequencers. The core advantage is that the decentralization level of the sequencer directly inherits from the decentralization of L1 validators, eliminating the need for a separate decentralized sequencer mechanism.

However, Based Rollup faces the challenge of finality delay—after ordering, blocks require waiting for block production and confirmation, which is less ideal for low-latency user interactions. Community proposals suggest combining pre-confirmation mechanisms with Based Rollup, aiming to provide strong protocol-level confirmation signals within 15 to 30 seconds.

Additionally, native Rollup pre-compile proposals are also under active development. Vitalik has indicated that the timeline for Ethereum L1 fully adopting ZK proofs and the integration of native Rollup pre-compiles is aligning, opening pathways to address the fragmentation caused by each L2 maintaining its own specialized proof systems. In the future, Rollups could leverage shared infrastructure for proof verification, avoiding high-cost, bespoke audit chains.

The Next Steps in Ethereum’s Upgrade Path Post-Glamsterdam and Hegotá

After completing Glamsterdam and Hegotá, Ethereum’s roadmap will enter a new phase called Strawmap. The Ethereum Foundation’s Protocol Cluster has seen leadership changes, with focus areas expanding to zkVM proofs, post-quantum cryptography coordination, zkEVM development, and protocol-level security at the trillion-dollar scale.

Strawmap is expected to continue the pattern of approximately two hard forks per year, with plans for seven forks before 2029. This indicates a move toward normalized, rapid iterative development—each fork no longer requiring the accumulation of numerous widely debated feature proposals, but progressing in an orderly, controlled manner, reducing the engineering risks associated with “big-bang” upgrades.

However, some EIPs from Glamsterdam, such as EIP-8237, have been deferred to later forks. Additionally, the governance issues surrounding Layer 2 decentralization remain unresolved; some L2s driven by commercial considerations may remain at Stage 1 for the foreseeable future. This underscores that, even as protocol-level technology advances, the decentralization of Layer 2 ultimately depends on balancing business models and protocol development.

Summary

Ethereum’s 2026 upgrade trajectory is at a clear inflection point: after three major upgrades—Fusaka (data availability), Glamsterdam (throughput and MEV governance), and Hegotá (cleanup and hardening)—its scalability capacity has significantly surpassed the initial bounds set by the 2020 Rollup-centric roadmap. Yet, the progress of Layer 2 moving toward full Stage 2 decentralization remains “slower and more challenging than expected.” Centralized sequencers, delayed fraud and validity proofs, and cross-chain fragmentation via multi-signature bridges constitute the three most intractable issues. The Glamsterdam devnet’s deployment of ePBS embedded in protocol and gas limit anchoring, Hegotá’s shift to “cleanup and hardening,” and the exploration of Based Rollup with pre-confirmation signals are all part of broader discussions aimed at addressing fragmentation with more cost-effective interoperability solutions.

Ultimately, the decentralization of Layer 2 is less about technical capability and more about balancing technical feasibility with economic incentives. Ethereum is adopting a pragmatic approach—acknowledging the coexistence of different stages within the ecosystem, and pushing the underlying layer to continue delivering verifiable progress through biannual forks, even if not all Layer 2s reach Stage 2 in the short term.

FAQ

Q: What is the current status of Glamsterdam devnet?

Glamsterdam-devnet-2 is live, with multi-client ePBS operating stably. End-to-end cross-client testing of the external Builder process has been completed, covering nearly all client implementations.

Q: What specific scalability milestones has Fusaka achieved?

Fusaka was activated on December 3, 2025, introducing PeerDAS (EIP-7594), embedding data availability sampling into the protocol. This allows Layer 2 to theoretically increase data availability capacity by about 8 times and significantly reduces node bandwidth requirements. The mainnet gas limit has been raised to approximately 60 million units.

Q: Why did Hegotá shift from scalability to “cleanup and hardening”?

Because after Fusaka and Glamsterdam, Ethereum’s L1 scalability has far exceeded initial expectations. Hegotá’s focus shifted to protocol-level hardening features like FOCIL for censorship resistance and account abstraction, returning the emphasis from throughput to security and decentralization.

Q: What are Based Rollup and pre-confirmation mechanisms?

Based Rollup assigns block ordering rights to Ethereum L1 validators instead of independent sequencers, inheriting decentralization from L1. Combining this with pre-confirmation mechanisms aims to provide predictable, fast confirmations within 15–30 seconds, addressing sequencer centralization and cross-rollup composability issues.

Q: How many stages is Layer 2 decentralization divided into?

L2BEAT categorizes stages as: Stage 0 (fully centralized), Stage 1 (limited multi-sig governance reliance), and Stage 2 (fully decentralized, relying solely on code and cryptography without backdoors). As of early 2026, most Layer 2s remain at Stage 0 or Stage 1, progressing more slowly than expected.