Almost none of that money goes towards improving Wikipedia. A sizeable chunk will go towards unpopular software projects like Flow[1] and costly rebranding efforts against the community's wishes.[2] The article mentions an alternative, the Wikipedia Reward Board,[3] where the money (or other reward) would actually go to the person (or people) doing the editing work. Not every Wikimedia project has a reward board, but for the ones that do, you can make your monetary contributions go a lot further by using that instead of donating to the WMF.
I would love for good contributors to be rewarded, but I'm not sure how you do that without people gaming the system and creating worse editor politics than there are now.
However, I am sure what happens if you don't pay the electricity and networking for the servers, or have a team of lawyers protecting the overall effort. Despite it's flaws, the current system is working well by my standards.
Not much of their budget is going toward hosting though -- in FY 18-19 it was ~$2.3M of their ~$91.4M expenses [1].
If we look back to 2005, Wikipedia was already one of the most popular websites on the internet, but their expenses were only $716k [2]. They were frugal back then!
I agree with the parent that the extra expenses are mostly superfluous. They have a ton of staff [3] and quite a few of their roles seem to be about promoting their brand or maximizing donations.
However, I am sure what happens if you don't pay the electricity and networking for the servers, or have a team of lawyers protecting the overall effort. Despite it's flaws, the current system is working well by my standards.