Fragmentation.
Blockchain is based on a peer to peer system. So each peer/node have the freedom to choose which version of the software to run. This makes it difficult to implement significant changes to the codebase because each peer may have different views on the changes.
I'd argue that this is both a boon and a bane.
Giving nodes the freedom to run different, possibly competing versions of the protocol provides an unprecedented form of autonomy to users. It keeps developers in check to not stray too far from what the users want and enables users to fork off should they perceive the original vision of a blockchain project to be compromised -- which is pretty much what happened with the Bitcoin Cash hardfork.
Of course it makes updates and reaching consensus regarding upgrades difficult, however I'd argue it's a price worth paying for making sure that no single entity gains too much power over a blockchain.
But when the 21m BTC supply was mined, miners will surely shift to a more mining-profitable coin that can lead to a lower Hashing power.
One of the side-effects of the rise of ASICs has been that Bitcoin is sharing its PoW scheme with only a handful of coins, signifcantly reducing the number of possible targets to alternatively point hashing power at. Nonetheless it is of course impossible to predict how the world and cryptocurrencies will look like a 100 years from now, given the latter still exist.