It's still a long time, even if we would say it could be critical 25 years from now, when Block rewards will start to get very low compared to now. But we had already phases, where fees made up 20-25% of block rewards. Yes, maybe on-chain fees will be a bit higher than today and for very cheap fees, lightning will be used. It's possible.
I expect, it'll be a constant process, where Bitcoin gets more valuable but even on-chain fees won't explode.
Remember SegWit, where Bitcoin's TPS was increased and more TX were fitted into a block without downside. TX costs and network congestion went down because of SegWit.
Maybe we will see some more similar updates to improve Bitcoin and maintain the affordability of on-chain transactions.
high fee's means people wont want to use bitcoin to even enter LN. they will instead exchange their fiat for a cheaper altcoin like litecoin. and then use that to enter LN to then play with millisats.
...
While I appreciate your dedicated time to reply here, I need to say that things are a bit more complex than outlined in your post. We just don't know yet how fees will work out and people have already said in the past (around 2017/2018) that Bitcoin will be impossible to use because of high fees.
Well...
What happened?
Nothing like that. Fees are cheaper than ever most of the last 5 years.
segwit increased TPS... WHEN?
moving upto 2017 bitcoin had the capacity of upto 4200 tx, and was in reality moving upto ~2500 average.. then segwit came in. and.............. 5 years later.. hmm.. flatline
The number of transactions that have happened is not a proof that the capacity didn't increase.
In fact, SegWit reduced network congestion and that's also a reason why we still have low fees of 0.10 cent today. There *could* be more transactions while tx fees wil remain low due to SegWit / Lightning.
Yeah, I remember how SegWit dropped the transaction prices and I guess something similar will have to be created so that more transactions fits into one block. Then maybe the block reward payed only by a transaction fee will be enough.
I didn't think that fiat will be worth anything in 120 years. I just made an example how would that look today in fiat value. But I guess one of the solutions is layer2 for small transactions. I'd write "we will see" but I guess we won't. Maybe our kids will see

1. they didnt drop the fee..
they instead made old transaction formats 4X more expensive. here its in the code. its a 4X increase of legacy
https://github.com/bitcoin/bitcoin/blob/master/src/consensus/consensus.h#L21https://github.com/bitcoin/bitcoin/blob/623745ca74cf3f54b474dac106f5802b7929503f/src/consensus/validation.h#L143see it says *4
imagine paypal had a 20cent fee per payment. and promised a 4x fee discount if people use paypal+.. then you find out what they actually done was make regular paypal initially 80cents to say that paypal+ is 4x cheaper (20cents).. and then weeks later they removed the fee formula that made prices 20cents and made it so paypal+ was $1-$3 average. meaning regular paypal(legacy) was $4-$16
What?

Legacy transactions didn't become 4x more expensive because SegWit was introduced?
Legacy transactions are just more expensive compared to SegWit transactions because SegWit transactions are cheaper because SegWit transaction have less block weight than legacy transactions.
Spinning this around that legacy transactions are now 4x more expensive does not make any sense...
In general, even legacy Transactions are benefiting from SegWit because it's much more unlikely to have a congested network.
also fee's in 2009-17 were sub 1cent-10cents.. fee's are now $1-$3
You can still get transactions of around 0.10$, I don't see a problem...

Even by using a legacy address, you can get close to 0.10$ fees...
But one of the benefits of SegWit was to make it possible to fit more transactions into one block. Shouldn't that cause transaction fees to be lower?
Exactly, SegWit reduces network congestion because more tx can be fitted into one block. Less network congestion = periods of high fees are less likely