The problem is that you need at the very least 1 on-chain transaction for every human participant, to get money into one wallet that they control. But BTC barely manages 5 TPS and ETH maybe 20. This means that BTC would need about 115 days to add 50M new people onto the block chain, and ETH could manage it in a month - only to have 1 wallet with some amount of coins in it (of course, assuming nothing else whatsoever is happening on the block chain in that time).

Only after you get past this hurdle do you even have access to a L2. Of course, you then need extra transactions on the L1 chain to gain access to the L2 - with LN at least, this scales even worse, as you need some fraction of the number of pairs of people to communicate - so some fraction of (50M)² transactions - which fraction depending on how centralized the network will be.

Then of course, L2s offer little in terms of guarantees for your money unless you can quickly and often close the channel and get the L1 guarantee.

So no, BTC and ETH fundamentally can't scale to a large number of people using them, regardless of how many trillions of transactions per second L2s can do. They are failed technologies for anything similar to their original goals.

It's barely even worth mentionibg that with 1 wallet per real person, you get worse privacy for your transactions than VISA offers you - as many have shown, for any kind of practical privacy you need many temporary wallets, bringing the total number of transactions required just to have a single Satoshi in those wallets to many times more.