If comparing Layer2 scaling to an interstellar journey, then zk-EVM is like a spaceship equipped with a superluminal engine, while the data layer is the fuel supply pipeline — it determines whether you can truly take off.

It is now December 2025. We have passed the peak of zk-EVM narrative's large-scale implementation, and are now in a period of calm. Over the past two years, the entire industry has cheered for a few seconds' improvement in zero-knowledge proof generation speed, and debated the definition of Ethereum equivalence for ages. But reality has dealt a harsh blow to developers: you finally built an engine capable of processing tens of thousands of transactions per second, only to find that most of the time, the engine is idling — because data supply can't keep up.

This is the most awkward predicament for Layer2 now: you have ample computing power, but are bottlenecked by data throughput.

Why is the halo of zk-EVM fading?

In early scaling efforts, we all thought that generating proofs (Proving) was the biggest challenge. But now? With hardware acceleration and parallel proof techniques gradually maturing, proof costs have come down. The real bottleneck is no longer "how to prove that these ten thousand transactions are valid," but "how to securely deliver the raw data of these ten thousand transactions to the prover with minimal latency."

Many Layer2 projects are now like driving a Ferrari through narrow alleys during rush hour. They are compatible with the Ethereum ecosystem, but when it comes to high-frequency finance or on-chain AI applications that require heavy data processing, they start to feel overwhelmed.