Some are also small servers run purely for the hell of it; if you keep the user count low it really doesn’t cost much to keep one running. These instances are of course more vulnerable to their owner being hit by a bus or whatever and not having shared their keys with anyone. Less so once account migration lands, and everyone’s gotten off their butts and upgraded their instances to that version.
(My instance, by example, currently has about 150 accounts, closed registration, and costs about twenty bucks a month to run. It’s also a few versions behind because I’ve been busy with other things.)
Every instance's own business model is often somewhat unique.
My primary instance at this point is on a domain I registered ages ago that I point to a SaaS host that runs and keeps up to date a growing number of small instances. I'm their customer and I expect them to charge me fairly based on my instance usage (CPU, bandwidth, et al) appropriately subsidized by their scale versus if I was booting my own VMs/containers and running it elsewhere.
If the worst happens, I expect I can move my business elsewhere, and given I control my domain itself, I can (use a data backup to bootstrap a new instance somewhere else and just point my domain DNS to the new instance).
There are a variety in business models among instances and the people hosting them.
Maybe they take donations, maybe you run your own instance and pay for the hosting. Not having shareholders that expect profits and growth probably takes a lot of the pressure off, too. All they have to do is not have burdensome costs or break even.
I'm just wondering how they are getting the capital to scale.
The decentralized nature helps the network scale.
And yeah, it's an open source project funded by donations. Plenty of those exist, and if it gets big enough, it'll get more contributors and donations.