Ethereum was cheaper than anticipated in 2020, and rollup decentralization was slower than promised in 2021. These two realities are compelled the ecosystem to rewrite what “a layer-2” is for.
Vitalik Buterin’s latest publish on Ethereum Analysis bluntly frames the shift: the unique imaginative and prescient of layer-2 (L2) blockchains as “branded shards” of Ethereum is now not viable, and the ecosystem requires a brand new path.
Nevertheless, this is not abandonment. As a substitute, it’s a re-tiering of expectations and a sharper definition of what several types of rollups are literally constructing.
The query now’s the brand new job description, for the reason that premise underlying the rollup-centric roadmap has weakened.
Stage 2 is scarce
L2BEAT gives the clearest framework for understanding rollup decentralization by its Phases system.
Stage 0 denotes that coaching wheels stay in place, with significant belief assumptions persisting.
Stage 1 represents partial decentralization with stronger escape hatches and proof ensures, however nonetheless significant improve or governance belief.
Stage 2 is the “no coaching wheels” milestone, during which crucial security properties are enforced by code quite than by discretionary actors.
The present distribution of worth secured throughout the L2 ecosystem signifies this. In accordance with L2BEAT’s rollup scaling abstract, roughly 91.5% of the listed worth sits in Stage 1 rollups, 8.5% in Stage 0, and roughly 0.01% in Stage 2.
The highest three rollups by worth account for roughly 71% of the overall, indicating that “Stage 2 progress” largely relies on the choices of the most important few tasks, quite than on what smaller experimental chains try.
The core blocker is whether or not the proof techniques may be overridden and whether or not upgrades face robust delays and constraints.
Improve discretion stays widespread among the many largest rollups, and transferring past it has confirmed slower and tougher than anticipated by the 2020-2021 optimism.
Some tasks have explicitly said that they might not want to proceed past Stage 1, citing not solely technical constraints associated to zkEVM security but in addition regulatory necessities that require absolute management.
That is a official product choice for sure buyer bases, but it surely clarifies that these chains will not be “scaling Ethereum” within the sense the rollup-centric roadmap initially meant.
| Mission | Stage | TVS ($) | Proof kind | Improve key / safety council current? | Notes |
|---|---|---|---|---|---|
| Arbitrum One | 1 | 16.16B | Optimistic | Sure | Emergency path can skip delays |
| Base Chain | 1 | 10.99B | Optimistic | Sure | Upgrades permitted by a number of events; no delay |
| OP Mainnet | 1 | 1.88B | Optimistic | Sure | Safety council instantaneous improve energy |
| Lighter | 0 (Appchain) | 1.27B | Validity | Sure | 21d delay, emergency can go to 0 |
| Starknet | 1 | 676.17M | Validity | Sure | Safety council can improve with no delay |
| Ink | 1 | 523.71M | Optimistic | Sure | Safety council + basis approvals; no common delay |
| Linea | 0 | 492.93M | Validity | Sure | Multisig can improve with no delay |
| ZKsync Period | 0 | 417.07M | Validity | Sure | Emergency board can bypass improve delays |
| Katana | 0 | 297.94M | Validity | Sure | safety council can take away the improve delay |
| Unichain | 1 | 168.81M | Optimistic | Sure | no exit window for normal upgrades; instantaneous powers |
Why the constraints modified
The Oct. 2, 2020, publish “A rollup-centric Ethereum roadmap” on the Fellowship of Ethereum Magicians laid out the unique thesis.
Fuel costs have been climbing, some purposes have been being compelled to close down, and the conclusion was that the ecosystem can be “all-in on rollups” for the close to and medium time period.
Base-layer scaling ought to prioritize knowledge capability for rollups, and customers would more and more dwell on L2.
Two laborious info have shifted since then. First, L1 is considerably cheaper at current. Etherscan reveals a seven-day common transaction payment of round $0.35 and gasoline snapshots within the fractions of a gwei.
On Jan. 16, Ethereum recorded an all-time excessive of two,885,524 transactions in a single day. The narrative is “busier and cheaper,” precisely the alternative of the 2020 disaster that motivated the rollup-centric roadmap.
Second, L1 execution capability is rising. Ethereum’s block gasoline restrict was raised to roughly 60 million after broad validator signaling in late 2025, up from the long-standing 30 million restrict.
At roughly 12-second blocks, 60 million gasoline interprets to roughly 5 million gasoline per second.
Aspirational neighborhood discussions have talked about targets as excessive as 180 million gasoline, which might characterize a threefold improve, although that continues to be directional quite than dedicated.
The clear interpretation: the 2020 premise that “L1 cannot scale for many customers” is weaker in at present’s payment regime. This creates room for L2s to be a spectrum of safety and sovereignty trade-offs quite than all being near-identical “shards” competing solely on worth.
L2s as a spectrum, not clones
Buterin’s proposed reframing treats L2s as occupying a full spectrum.
On one finish are chains backed by the complete religion and credit score of Ethereum, with distinctive properties, not simply EVM clones but in addition privacy-focused techniques, non-EVM execution environments, or ultra-low-latency sequencers.
On the different finish are choices with various ranges of Ethereum connectivity that customers and purposes can select based mostly on their particular wants.
The brand new minimal bar is simple: if you happen to deal with ETH or Ethereum-issued belongings, attain no less than Stage 1.
In any other case, you are a separate L1 with a bridge, and may name your self that. The differentiation bar is tougher: be the perfect at one thing aside from “low-cost EVM.”
Examples Buterin cites embrace privateness, effectivity specialised to a selected utility, actually excessive scaling past even an expanded L1, basically completely different designs for non-financial purposes resembling social or id techniques, ultra-low-latency sequencing, or options resembling built-in oracles or decentralized dispute decision that are not computationally verifiable.
The mechanism that may facilitate that is nonetheless underneath investigation. A “native rollup precompile” would allow Ethereum to confirm a typical zkEVM proof inside the protocol.
For rollups which might be “EVM plus extras,” this implies the canonical EVM verification happens trustlessly on the protocol stage, and the rollup solely must show its customized extensions individually.
This might allow stronger interoperability and pave the best way for synchronous composability, during which contracts throughout completely different rollups can work together inside the identical transaction. But, it stays a analysis trajectory, not a deployed characteristic.
The Jan. 16 publish “Combining preconfirmations with based mostly rollups for synchronous composability” and the Feb. 2 publish “Synchronous composability between rollups through realtime proving” lay out the design house however do not characterize shipped protocol modifications.
Three buckets rising
If this reframing takes maintain, count on rollups to separate into clearer classes.
The primary bucket is Stage 2-chasing settlement rollups that maximize Ethereum safety inheritance.
These tasks goal to realize code-enforced ensures with minimal discretionary governance, treating “scaling Ethereum” as their core mandate.
The second bucket is regulated or managed execution environments.
These optimize for compliance, permissioning, or particular institutional necessities. They could by no means progress past Stage 1 by design, and they need to market that management truthfully as a characteristic quite than pretending to supply full decentralization.
The third bucket is specialised chains optimized for latency, privateness, app-specific execution, or non-financial use instances.
Privateness rollups utilizing zkProofs to cover transaction particulars, ultra-low-latency sequencers for buying and selling purposes, or social and id techniques with basically completely different state fashions all fall inside this class.
These do not must be EVM-compatible and even monetary to justify their existence, they should present worth that their customers cannot get elsewhere.
Tasks resembling Arbitrum One, Optimism, Base, zkSync Period, and Starknet will every must determine which class they’re pursuing. The ecosystem is giant sufficient to assist all three, however the assumption that each L2 performs the identical operate is fading.
What modifications for customers and builders
For customers, the burden shifts to understanding ensures. Escape hatches, improve delays, proof techniques, and censorship resistance turn into product differentiators quite than assumed properties.
Wallets and interfaces might want to label belief assumptions extra explicitly, and the L2BEAT Phases framework goals to make these assumptions legible.
For builders, “low-cost EVM” is commoditized. Differentiation strikes to privateness and customized digital machines, ultra-low-latency sequencing, app-specific throughput optimizations, non-financial purposes in social, id, or AI contexts, or compliance and permissioning as an express product, with out claiming it is “Ethereum scaling.”
For the broader market narrative, count on a louder debate about whether or not L2s “inherit Ethereum safety” in apply quite than as an aspiration.
The critique is already a speaking level amongst rival L1 proponents, and the ecosystem’s acknowledgment that many giant rollups stay at Stage 1 with discretionary governance provides that…
