Ethereum is going to need “infinite capacity” to serve its “World Computer” ambitions, Joe Lubin argued at a recent talk. His conclusion was blunt. L2s are not optional.
Lubin’s bigger bet is a longer runway. He suggested Ethereum could evolve into a fully zero-knowledge proof based protocol in roughly 3 to 5 years, if the ecosystem keeps converging on the right architecture. That is a roadmap claim, not a guarantee. But it frames the direction of travel more clearly than the usual “zk is coming” refrain.
Why Lubin keeps pointing to L2s
Lubin tied Ethereum’s scaling problem to a simple mismatch. Demand for computation does not cap out neatly. Meanwhile, the base layer has finite capacity.
His prescription is the familiar one, but with extra emphasis on necessity. “Ethereum needs infinite capacity… making L2s necessary,” Lubin said. In other words, rollups and other offchain execution models are the immediate pressure valve, even if the endgame is a different proving model.
For users and operators, that matters because it changes what “scaling progress” really means. The network may not feel faster at layer 1 in the way people expect from past L1 upgrades. Instead, capacity moves through L2 throughput and settlement properties.
The 3–5 year zk endgame claim
The sketch Lubin offered goes further than scaling alone. He said Ethereum could become a fully zero-knowledge proof based protocol in 3 to 5 years.
That is not just a technical preference. A fully zk-based design implies core state transition logic and verification patterns that lean heavily on proof systems, not just selective components. The practical consequence is that it would require deep integration across the stack. That includes how transactions are formatted, how the network verifies validity, and how clients implement those rules.
The timeframe also raises the bar for execution. A 3 to 5 year window is tight if the goal is broad migration of protocol behavior, not just incremental rollout of zk features.
What this means for infrastructure reality
Lubin’s comments come from a builder perspective. Still, protocol reporters have to ask the mundane questions. Who runs the machinery. How many independent implementations exist. What happens during edge-case failures.
A shift toward a fully zk-based protocol would not just be about cryptography. It would also be about engineering reality on validator and client sides. Proof generation and verification costs, network propagation patterns, and operational tooling all need to line up. The more complex the verification path, the more sensitive the system can become to client diversity and outage behavior.
And there is the ecosystem question. Even if Ethereum’s base layer trends toward zk verification, L2s already coordinate much of today’s throughput. That means L2 designs and proving strategies become the proving ground for what “fully zk” can mean in practice.
The trade to watch: world-computer scale vs base-layer limits
Lubin’s framing is consistent with a market reality. Everyone wants “world computer” throughput. The base layer cannot just brute-force that demand without paying for it in congestion and resource costs.
So the real question is whether the ecosystem can route that demand through a path that preserves composability, security guarantees, and verifiable execution. L2s do some of that now. A future fully zk-based protocol would aim to make the verification story more uniform.
But until there’s shipped protocol detail beyond high-level claims, the safest stance is skepticism with a checklist. Track which zk components actually reach production. Track client and proving system maturity. Track operational stability.
What Lubin said, in one place
| Claim | What Lubin tied it to | Direct source text |
|---|---|---|
| Ethereum may become a fully zero-knowledge proof based protocol in 3 to 5 years | Long-term direction after L2 adoption | “Ethereum could become a fully zero-knowledge proof based protocol in 3 to 5 years” (Lubin, via The Block) |
| L2s are necessary | Ethereum needs unbounded capacity for “World Computer” demands | “Ethereum needs infinite capacity… making L2s necessary” (Lubin, via The Block) |
Source: The Block, reporting Joe Lubin’s remarks.