The history of last winter tells that the 1 MB block size may be better for miners than larger sizes because that can force a quite big rise in fees if the number of transactions grow enough.
So if you base your calculations on the assumptions that the number of transactions in 100 years will be at levels comparable the ones we have today and the price will also stay on this level, you are wrong. The price has to rise a lot (buzz words: deflationary coin, scarcity), the number of transactions (on chain and off chain should grow, maybe even exponentially)...

Then there's the difficulty. That may make my calculations wrong, but Bitcoin had in history moments when mining was not worth it anymore to certain categories. They stopped their gear and maybe the difficulty didn't rise that fast.  This could also happen in the future too if the price continues to drop.