The announcement of KaOS 2026.02 seems to have created a lot of questions regarding the possible move away from systemd. Since that announcement did not go into detail, this post will try and clarify the situation.
Reasons for KaOS existence
The simple reason why KaOS exists is that it is trying to provide things no other distribution does. Any time a new distribution is announced, the question always comes, why another one? A distribution like KaOS did not start building roughly 2200 packages from scratch, sets up all infrastructure for repositories, code, website, forum, build-system, ISO releases, if another distribution fulfilled the need. Moving away from proprietary Operating Systems to open source options (Linux-based, BSD based, Solaris based) is about wanting freedom and choice in almost all cases. But should any such Operating System or Distribution not make some choices of what it believes is the best fit? KaOS sees a lack of focus in that respect. To create the highest quality Distribution possible, there needs to be a focus to make sure the user gets the best possible for whatever choice they made. It simply is not possible to package any and all to work perfect for every Desktop Environment or Toolkit, thus the focus on one Desktop Environment, one Toolkit.
When KaOS started in 2013, the idea of symlinking a whole distribution to /usr had only come into play, just two mainstream distributions had adopted it, but the biggest once rejected it that time. KaOS choose to stay with the traditional approach of having a separate /usr partition, seeing it as a cleaner option then symlinking a whole distribution together.
Around 2010/2012 it was not clear yet whether AUFS or OverlayFS was going to be part of the kernel. Since there was already about 3 years of experience with AUFS, KaOS decided in 2013 to use the AUFS filesystem for its overlayfs implementation. To this day, it has it been easy to implement, has created a very smooth option for user to select non-free drivers right from the start of a Live system and is well maintained. So, KaOS has always chosen AUFS over Overlayfs, even though the latter has become part of the kernel.
The above are all core reasons why even have a small, independent distribution, if none of those are important to anyone, why even have KaOS?
Changes in systemd
In 2013, systemd was the modern init system, fully modular, fitting very well for KaOS. It has been a very good init system for years here, but with the announcement of systemd 254 in July of 2023 (last version to fully support the split /usr setup), it became clear, the future of KaOS as is, was endangered. Searches for how to approach this problem went on for a good two years, hoping for a simple solution. But when AUFS support also stopped with newer versions of systemd, it was obvious that KaOS would need to find a new init system. Or it would come to a point there were very few reasons left to keep this distribution.
If just following the mandates that systemd puts out (and used by most/all mainstream distributions), there is no need for a distribution that always had used its own ideas and path. If you are just going to do what everyone else is doing, no need to put all the effort in maintaining code, packages and infrastructure, just use one of the main distributions.
Trials to move to Dinit and how that affected Plasma
By the middle of 2025, trials were started to move to Dinit as new init system. But as expected, this was not a simple task, and would take months of work to implement for a smooth transition for all users. Since the developers of the Plasma desktop started making more and more commits that emphasized the integration of systemd in Plasma (examples, drkonqi started making systemd a hard requirement early 2025, plasma-login-manager being systemd only), it started to make more sense to first see if there was another option for KaOS then the Plasma Desktop. Since replacing the top layer is far easier than replacing the foundation of a system.
KDE Plasma and KaOS
Make no mistake about it, the Plasma desktop is an incredible product, and considered by far the best desktop environment here. There would be no way of looking for any alternative if a move away from systemd was not mandatory here. But if you are going through the effort of completely redoing the base of a distribution, you need to also base yourself on something that is fully supported. Non-systemd systems can’t be as well supported by Plasma as systemd based ones. Not a case of just hard requirements, but also a case of moving parts to legacy. The Plasma sessions scripts for example (the part that runs right after the login manager), split out a non-systemd section and is now labeled as legacy. And as with X11, moving parts to legacy most probably means it will not be supported in the future. Not being able to use drkonqi for bug reports, will make those reports less useful, so will make it harder for KDE developers to support reports from non-systemd distributions.
The niri/Noctalia Future
To stay true to its initial goals, one DE, one Toolkit, clean filesystem setup, the combination of Niri and Noctalia Shell have shown to be great option for KaOS. It still is a Qt (QML) based shell, applications can still focus on Qt/KDE options, an initial install is still GTK free. Will it be good enough to fully replace Plasma, and give KaOS a future? Or is it too limited and a full Desktop Environment is mandatory, thus making KaOS obsolete…
If you have comments or questions regarding this article, please visit the KaOS forum and share your thoughts.