Doing a bit of Googling, and I found these posts:

I've only scanned the first three ... they provide sufficient info. The rest the posts, I suspect not as good. Who knows.

The tl;dr: is that yes -- over the next 120 or whatever years, fees must rise to a sufficient level to maintain a sufficient level of protection against a 51% attack.

Some nocoiners, shitcoiners, and such make (in my opinion) this risk a much bigger deal than it actually is (to try and make their sthitcoin look less bad) or try to shill some "solution" (e.g., changing to some other type of mining, or proof-of-stake, whatever).

If bitcoin continues towards the path of mass adoption, on-chain transactions will be priced out for most anything but settlement transactions and channel open / close transactions. So a robust fee-marker may certainly return and thus fee revenues to miners would increase.

One of the problems is people think bitcoin mining today costs $40M per day (because that's what it costs to run today's ~200 EH/s of mining capacity. (The $40M number is based their revenues today ... 900 BTC/day X $47K BTC/USD, so over $40M/day).

But consider a bank whose vault walls are 24-inch thick steel. Is a bank with 12-inch thick steel walls only half as safe? If the main threat is a cutting torch over a 3-day weekend, then maybe no -- a 12-inch thick steel wall might be just as secure as a 24-inch thick steel wall.

That's the case with bitcoin. The reason it grew to 200 EH/s of mining is because that's what the current mining rewards will support. Hashrate follows price. If the bitcoin price doubles, then the mining hashrate essentially doubles as well (all else being equal). So after the next halving, the block subsidy revenue to miners will drop by half. That's a known fact. Let's say hashrate then drops by half (it won't, but hypothetically ...). Could bitcoin survive on just 100 EH/s? It did just fine on 100 EH/s a couple years ago. Why would that be highly risky once again? I suppose if there was 100 EH/s of discarded (then obsolete) ASICs that fell into the wrong hands, the technical risk of a 51% attack could increase. Economic game theory says that won't happen, but ... moving on.

Let's say a 51% attack was successful. There's always the nuclear option ... change the mining algo overnight to a new algo, where with no ASIC hardware existing, we start the race again, from millions of millions of CPUs, then GPUs, then genA ASICs, then genB ASICs, and so on.

As far as the future ... there's some people (such as Messari's CEO) who remain convinced the 21M cap won't stay forever:

Of course, he's 100% wrong, because what he proposes would not be bitcoin, it would be something else. And because it would be contentious, we'ld see a contentious hard fork (if pushed), ... and then we'ld have Bitcoin and BitcoinInflate (or whatever name it gets). Tell me if you've heard this one before, and I'll stop. Actually, ... I'm going to stop.

The final word: If we knew the future, bitcoin would either be $2M BTC/USD or it would be $0. Because we don't know the future, BTC/USD falls somewhere in the middle. We don't know, for sure, if bitcoin will gain mass adoption. We don't know, for sure, if the number of on-chain fees will rise again and a fee market return. We don't know what hashrate level is required to be sufficient protection against a 51% attack. We don't know ... price, etc, etc ,etc.

What we can say is, ... people today trust the plan enough to value each of the ~19M BTC that exist at about $47K each. That's the only thing we do know, today, with absolute certainty.