Where do Base and Arbitrum win? Not technology—it's the ecosystem.

Yesterday’s article shared a batch of old projects that are pinning their last hopes of survival on migrating their ecosystems.

The target ecosystems for these migrations almost all converge on two Ethereum Layer 2 scaling solutions: BASE and Arbitrum.

During the infrastructure boom of Layer 2 scaling in the past few years, a large number of Layer 2 solutions emerged. But today, most of these Layer 2s have long since become “ghost towns.”

If we look at system-deposited TVL, according to defillama data, the top two are BASE and Arbitrum. Even the Layer 2 in third place, OP Mainnet, has only a small fraction of Arbitrum’s deposited TVL.

Other Layer 2 scaling solutions are nearly negligible.

So it can be said that the ones most likely to survive from that previous Layer 2 infrastructure boom will ultimately be these two—or, optimistic for a stretch, not more than five.

In that infrastructure boom, there was a very clear characteristic in how Layer 2s were built: you first “build the nest,” then “attract the birds.”

As for what kind of “birds” you can attract after building the nest—and whether you can attract the “birds” you expect—nobody knows. So when each team builds its own Layer 2, what features to prioritize on the Layer 2—performance-first, security-first, or usability-first—is, to a large extent, entirely a matter of the team’s own “subjective” judgment.

That’s why we saw Layer 2s built on zero-knowledge proofs (e.g., StarkNet, zkSync), Layer 2s built on Fault Tolerant (e.g., OP Mainnet, Arbitrum), and Layer 2s that primarily pitch transaction performance (Scroll, MegaETH).

And can those “nests” built on “subjective judgment” attract “birds”?

After years of market testing, it’s fair to say that most have failed.

The main reason, I think, isn’t that the technology wasn’t advanced enough—but that they lost the ability to truly understand market demand and user needs.

In that situation, even the most advanced technology doesn’t help.

Among the Layer 2s mentioned above, the easiest technically to implement are the ones based on Fault Tolerant. Ironically, today’s most thriving BASE and Arbitrum are also built on this simplest technology.

Why did the teams at the time lose true insight into market demand and user needs?

I believe it was a historical misstep of the era, not necessarily that the teams intentionally ignored it.

Because in that period, the most flourishing part of the entire crypto ecosystem was DeFi. And when it comes to DeFi, we will surely think about capital security and transaction speed. Once you think about capital security and transaction speed, you inevitably think that you should build a Layer 2 using the safest and highest-performance technology.

Where was this idea wrong?

In the historical context back then, I don’t think it was wrong. But two situations happened that people couldn’t have imagined at the time:

First, no one could foresee that Ethereum’s mainnet would later undergo large-scale scaling.

This scaling greatly reduced Layer 2 fees and greatly increased system performance. It means that even if Layer 2 performance is still not ideal today, it’s more than enough to handle current transaction scenarios. So the performance differences brought by technology between Layer 2s are hard to translate into real advantages.

Second, the mass high-frequency transactions in the offline (off-chain) world that people had envisioned didn’t actually get put on-chain that quickly. So even using the simplest technology to handle today’s transaction scenarios is already sufficient.

Under these circumstances, which ecosystem can attract more users and stay more active in demand matters far more than which Layer 2 has higher performance.

Also, according to Vitalik’s roadmap, Ethereum will continue to scale in the future and will also introduce ZK technology into the mainnet. That further weakens any differences Layer 2s might create purely in transaction performance.

We might even see a scenario like this:

In the future, for ordinary transactions, even if there are differences between the mainnet and Layer 2s in both fees and transaction speed, those differences won’t be so large that people can’t tolerate them.

I especially want to mention one case: MegaETH.

I remember this Ethereum Layer 2 in particular because I studied its whitepaper in detail at the time. It emphasized that it could achieve TPS at the million level through technology—once that technology was realized, it could essentially recreate on-chain something with performance close to a centralized stock exchange.

Now it has been built, and according to the official claims, its performance is also quite excellent.

The awkward part is that there isn’t really any such trading venue running on top of it. And over the past few days, AAVE withdrew funds from its ecosystem, which immediately knocked down its on-chain deposited TVL by 60%.

This Layer 2, which claims top-tier performance, is rapidly becoming a “ghost town.”

By contrast:

Robinhood, the traditional centralized exchange, chose a (relatively) simplest technical stack—Fault Tolerant—to build its Layer 2. After going live just a few days ago, it kept setting great records in user interest, activity, and trading volume.

Why?

Because of its user base, marketing approach, and market strategy.

Whether those achievements can be sustained still needs to be observed, but I believe that even if it can’t continue, the reason won’t be that it gets stuck on technology and performance.

In other words, its technology and performance can’t compare to MegaETH, but those two shortcomings will very likely not prevent it from continuing to expand and build out its own ecosystem in the future.

So the way we think about building Layer 2s has fundamentally changed.

In the future, if we build a new Layer 2, the primary goal must be business scenarios and user activity. Technology isn’t unimportant, but it may move to a relatively secondary position—as long as it’s sufficient or has a bit of redundancy.