I'll just copy & paste my recent comment on it. I don't think we'll ever need it but...

There's a (sort of) solution to this tail emission discussion, not mentioned in the article.

We will have to hard fork Bitcoin in the future (the year 2038 problem), don't panic, we have time until 2100-ish. With this hard fork, we might do more changes at once, like increase the block size, something we still don't need. I'm sure there will be other issues waiting to be sorted.

Anyway, what we can also do, is to go smaller units than satoshis. We already do use millisats (three decimal spaces) on the LN, we might go much further.

Now, the last halving in 2140-ish will dictate the block reward at 1 satoshi. What if it won't be the last halving but we will have a following cycles, rewarding miners 0.5 sahoshi, the next 0.25 sat, etc?

This way, we won't need to implement the tail emission and the 21 million coin will still be intact.

Note: there's this limiting factor, the max value of a 64 bit integer (18,446,744,073,709,551,615) but I'm sure we'll find a way around it too.

You can't. It would be a hard fork. Part of full nodes since the buffer overflow the resulted in a community supported 24 hour roll-back is that no node will accept a block which increases the Bitcoin supply past 21mil