Pursuing decentralized, quick-to-deploy, user-owned cloud solutions is a matter of privacy and resilience: centralized cloud solutions mean that personal (meta)data and content of private conversations of hundreds/thousands/millions of people are processed by a single entity, which constitutes a huge single point of failure, both from a privacy and from a service availability point of view.
As smart working in the era of COVID-19 is teaching to all of us (with problems of overcrowding, insufficient bandwidth, traffic bottlenecks between clients and centralized servers, etc.), in case of emergency, decentralized solutions are proven to be more resilient and effective to keep services available to all (or at least to most). In the end, the Internet itself was originally designed as a fully decentralized network to the very same purpose (to be resilient also in case of war).
Because it is a stable and reliable open source videoconferencing solution that can be deployed in minutes at any time on an inexpensive server (no special horsepower requirements; no need for any particular data storage, configuration, interfacing with external applications etc.). And it is very easy to use on client side, too.
Someone may say “why can’t I use one of the dozens of Jitsi instances already deployed and available for everybody?”. You can, but you have to trust the host: it would be just like using any commercial alternative. Open source on its own does mean nothing if you cannot look under the hood: for example, the host could activate chat logging or video recording without your knowledge. Besides, a single host serving a lot of third party conversations may suffer from bandwidth/system overload and is more likely to be targeted by malicious hackers.
REMEMBER “There’s no cloud… just other people’s computers”