Lightning isn't even a good solution for most diehard bitcoin users. It's a failed project.
It would take 27 years to onboard every internet user to the lightning network unless you start adding level 3 aggregators and then at that point you lose all the benefit of it being on chain at all.
It would take almost 2 years just to onboard every American assuming during that time there were zero other bitcoin transactions. Then you need to add the fees for the on and off ramps to the individual transaction fees to get the real cost per transaction noting that these would go up quite a bit as the competition between lightning and non-lightning uses of the transaction space would drive prices higher.
The throughput is arbitrarialy limited by bitcoin's current block size, which hasn't been increased since satoshi's era.
Most cryptocurrencies have an adaptive block-size mechanism which allows the blocks to grow to a reasonable size which could facilitate such an onboarding of users. So it isn't a technical problem, it is just a question of bitcoin's current leadership, which is controlled by companies like blockstream.
People have been debating the blocksize for a very long time now and there doesn't seem to be any large desire to change it so while the ability to increase it exists changing anything that fundamental about bitcoin seems to be a non-starter and while that is true lightning is pointless as a solution for the masses.
Even if you increase the block size 100x though you're still not improving the numbers much since my very generous numbers ignore activity outside of lightning and assume a single on chain transaction for every user and a perfect network.
Maybe your understanding of lightning is wrong here. Yes you open channels, and transactions in lightning need open channels, but you do NOT NEED to open channels for specific transactions. You open channels once, and transact over them for years. I run a lightning node with more than 15 channels (each to different lightning nodes) that are all older than 1 year (I route payments, so I have way more channels than needed). You can batch-open channels, i.e. I could have opened all my 15 channels with a single on-chain transaction. Taproot update would make those "commitment transactions" onchain way smaller (in byte) than needed to in the past.
Once channels are open, the users on the lightning network can transact back and forth without any new channel opens/closure and thus no on-chain settlement. Hence: Throughput in lightning is not at all limited on the bitcoin transaction throughput.
You mistake is, you overestimate the amount of people who want to be self custodial. You don't need to onboard every human being in the world on-chain.
Given the US example that would be several years in the absolute best for lightning case to onboard even 5% of individuals. Lightning is doomed from the start.
And if you don't care about self custody then the overhead of using a blockchain is a waste.
It is not black/white. It is okay to have the freedom to become self-custodial anytime, but not everybody needs to transact in self-custody all the time.
Taproot was another major step that enables lightning upgrades in future versions (such as zero-fee channel opens) that is barely discussed. The number of X years for onboarding Y amount of people is not accurate, as it disregards all major developments of the last 5 years.
You need enough users for providers to bother making it an option to pay with so it really is black and white. You might get a few niche providers offering it as a payment method without a critical mass of users but most companies aren't going to invest time and effort into implementing a payment system a tiny percentage of users have access too and if I need to trade money with my friends the low % means that in the vast majority of cases they aren't going to have lightning available either.
It would take 27 years to onboard every internet user to the lightning network unless you start adding level 3 aggregators and then at that point you lose all the benefit of it being on chain at all.
It would take almost 2 years just to onboard every American assuming during that time there were zero other bitcoin transactions. Then you need to add the fees for the on and off ramps to the individual transaction fees to get the real cost per transaction noting that these would go up quite a bit as the competition between lightning and non-lightning uses of the transaction space would drive prices higher.