If Ethereum is going to rely completely on L2s for scalability, then they (L2s) should start shifting from a monolithic architecture to a more modular one

The major obstacle that L2s are facing right now in terms of hyper-scalability is the fact that they’re built using a monolithic architecture. This means that settlement, execution and data availability are all still wrapped in the same layer.

This works for now, but if we want a more broader form of Ethereum blockchain adoption then this alone won’t do.
This approach that L2s currently have will soon get them clogged up.

Now some L2s have found a “solution” for this by moving data availability (which is by far the most major issue that Ethereum faces) to another layer. The only problem is that these layers are nore centralized and are run by third parties, so obviously security is compromised.

A modular approach however has far better perks and, more importantly, provides more throughput.

As of now, no L2 has still built something like this but Mantle Network is in the works of developing this model

This will allow them to build decentralized data layers that allow re-staking. This means that anyone who has staked ETH on layer-1 can re-stake a pegged asset and provide security to Mantle’s data availability layer.

If all L2s follow suit then Ethereum’s ecosystem will be ready for mainstream adoption.
Output would go through the roof all while security isnt being compromised.

But using the current model, we can only go so far before the system gets clogged.
The current model can support a couple hundred thousand users but I’m talking about millions of daily users.

Sharding should also help with this but again, if everything is being processed in a monolithic and sequential manner, then we can only go so far.