Here's an excerpt from my August 2022 article on Lightning that has relevance for today:
https://www.lynalden.com/lightning-network/

Criticism 6) It Has a Scaling Ceiling for Self-Custodial Users

Lightning greatly increases the transaction volume that is possible on the Bitcoin network. However, opening and closing a Lightning channel still requires an on-chain transaction, which means that in its current form, the Bitcoin/Lightning stack still can’t scale to billions of people using it self-custodially.

Specifically, there are block space limits to how many people can use it fully self-custodially on a regular basis, unless certain base layer forks allow for more throughput.

For any network, there are inescapable technical trade-offs. To ensure the widespread auditability and immutability of the base layer, there are some constraints that are hard to overcome.

I view many other blockchain designers as trying to over-engineer their systems. Any solution needs to have product-market fit. Not everybody wants a fully self-custodial experience. Some people want the convenience of using a custodial service of some sort. Bitcoin/Lightning gives optionality to people around the world, but people can see fit to use whichever portion of the stack that they want.

To quantify it, the Bitcoin/Lightning stack can be semi-regularly used by tens of millions of people self-custodially (or more than that over time if many of them are just holding it in cold storage). Custodial services can scale that to higher numbers. For example, all of the tens of millions of accounts on Cash App technically have access to the Lightning network, through nodes and channels operated by Cash App. The same is true for people on Strike, River, and similar types of apps.

At the current time, the Bitcoin network is being criticized by some opponents for low fees and thus supposedly low long-term censorship resistance as the block subsidy winds down (meaning there is not overwhelming demand for its block space at the current time, which if that state were to persist indefinitely could eventually result in a low cost to control over half of the mining share), while it is simultaneously being criticized for not being able to scale self-custodially to everyone in the world (meaning its block space is not nearly big enough to fulfill such enormous potential demand). These are mostly mutually exclusive concerns.

If the combination of the Bitcoin/Lightning stack eventually reaches severe growing pains against the number of people that want to interact with it fully self-custodially (a good thing), then there are additional areas of development that can increase its scaling potential, via ways to allow more users to share a given channel, which are beyond the technical scope of this article.

On the other hand, if the network doesn’t grow much and its block space does not increase in value (a bad thing), then its scaling limitations are a non-issue.

At the current time, the Bitcoin/Lightning stack provides tremendous scaling potential compared to the number of people that currently use the network. The network doesn’t need to overbuild for market conditions that don’t exist yet, although of course it’s good for developers to be thinking about long-term scaling options.

As the saying goes, “necessity is the mother of invention”, and if/when the network encounters persistently high base layer fees, tons of base layer transactions being used to open lightning channels, and an inability to onboard all of the users that want to onboard to the network self-custodially, then that would spark more interest in developing further scaling solutions, including the possibility of new broad-consensus soft forks and other changes.