I doubt bitcoin will last in the long term, 10+ years. Most alt coins will tank long before btc though, every litecoin clone as well as litecoin are doomed imo.
The Bitcoin network costs ~$150M a year to run. Either way you pay for it. Users (collectively) pay for the cost of securing the network by inflation or by transaction fees. The cost is real. One could hard fork Bitcoin so that it has continual block subsidy at 12.5 BTC after the next subsidy cut, and eliminate all fees (possibly using a tx POW to prevent denial of service). This wouldn't make Bitcoin free. The network has a real operating cost and users would pay for it in the form of inflation.
It is also a fallacy to say fees have to be "too high" in order to support the network. Not necessarily, it all depends on tx volume. At $150M a year that works out to $2853 per block. To keep the current level of network strength miners need to be compensated ~$2853 per block. You could have $2853 in subsidy (inflation), $2853 in fees or more likely some combination of both. Miners don't care. Either way the miners are supported by the same amount of annual revenue.
The larger the network the lower fees can be. Assuming no block subsidy the fee per tx would be
@ 1 tps = $4.76 per tx
@ 7 tps = $0.68 per tx
@ 50 tps = $0.10 per tx
@ 5,000 tps = $0.01 per tx
If you want to look at it terms of Bitcoins. Today miners are compensated an average of 0.12 BTC (120 mBTC) per transaction by a combination of subsidy and fees. That is relatively high but the marginal cost per tx is very low (maybe 0.01 mBTC) so the network can support significantly higher volume without significantly higher cost. How high? Well that is a matter of some debate but at a minimum 10x to 100x is pretty conservative. At 10x current volume that would put the cost per tx @ 12 mBTC and would be lower than many other payment systems. At 100x it is more like 1 mBTC and Bitcoin would be the lowest cost transaction system in the world.
Of course due to the relatively high subsidy the network is likely stronger than it needs to be (based on the value of economic tx it protects). So there is plenty of time before fees have to become even a majority of miner's revenue. It does show the genius of Satoshi system. The network has real cost and if users directly pay that real cost solely in tx fees the project would be nonviable from day one. Instead miners are subsidized to support the network while it grows. Your claim is only true in the network remains relatively small even after a couple decades. Then again if the network remains relatively small after a couple decades then Bitcoin likely is a failure. A high subsidies wouldn't change that fact.