Ubuntu, Manjaro, and the Linux Desktop I Thought I'd Left Behind

I have not switched back to Kubuntu yet.

I am still daily-driving Manjaro KDE on my ASUS Vivobook Pro 15, with Plasma 6.6.5, Wayland, btrfs, and a hybrid Intel Arc + NVIDIA RTX 3050 setup. But I have been running Kubuntu 26.04 LTS in VirtualBox, comparing the two side by side, and asking a question I did not expect to ask again after several happy years on Arch-based rolling release:

Did the distro I left quietly get good enough that the trade-offs flipped?

This post is my attempt to answer that honestly — with facts checked against current release notes, not nostalgia, and not a “winner” declaration.

A short history of my desktop Linux choices

I have been using Linux on the desktop for a long time. I started in the early Ubuntu era (back when “Ubuntu” still meant a fresh GNOME 2 desktop and a community that felt like it was building something new), stayed through the Unity years, moved to Kubuntu when Ubuntu’s default desktop direction stopped matching what I wanted, and eventually landed on Manjaro KDE.

The reasons for leaving Ubuntu-family distros back then were real:

• KDE on Kubuntu often lagged behind what KDE was shipping upstream.
• Kernels felt old on LTS unless you chased HWE stacks, PPAs, or third-party mainline tooling.
• PPAs accumulated for one app, one driver tweak, one toolchain — and became their own maintenance project.
• Snap was young and rough; many of us wanted .deb or native packages first.
• On laptops, hybrid graphics (Intel + NVIDIA) was frequently more painful on Ubuntu than we wanted to admit.

Manjaro solved a bundle of problems at once:

• One primary package manager (pacman) plus AUR access for everything else.
• Newer kernels and Mesa without hand-rolling.
• Newer Plasma on a schedule that felt closer to upstream KDE.
• mhwd for GPU driver profiles without reading three wiki pages on day one.

For years, that was a strong deal.

What actually changed in 2026

The surprise is not “Ubuntu is perfect now.” The surprise is how much of the old gap closed on a current LTS.

According to the Kubuntu 26.04 LTS release notes, Resolute Raccoon ships:

My Manjaro install today is in the same generation: Plasma 6.6.5, kernel 7.0.10, Wayland, Qt 6-based Breeze theming. In a VM, Kubuntu 26.04 reported Plasma 6.6.4, Qt 6.10.2, Frameworks 6.24.0, and kernel 7.0.0-22-generic.

That is not “Ubuntu is five years behind KDE anymore.” That is the same major era of the Linux desktop, with different packaging philosophy underneath.

A few corrections worth stating up front, because overselling helps nobody:

1. Ubuntu Desktop (GNOME) and Kubuntu are not the same product. Ubuntu 26.04’s GNOME session is Wayland-only in ways that do not apply identically to Kubuntu. Kubuntu still documents an optional X11 session package (plasma-session-x11) in the archive, but the team does not install or support it by default.
2. Kernel freshness on Ubuntu LTS is not magic. You get Linux 7.0 in 26.04 as part of the release. If you need kernels ahead of what Canonical ships on your LTS, you still use HWE paths, vendor stacks, or third-party mainline tooling — I have used Mainline on Ubuntu before myself. Manjaro’s rolling model is still simpler if “always newest kernel” is the goal.
3. Snap is part of the Ubuntu world whether you love it or not. On Kubuntu 26.04, Firefox is delivered as a Snap. That is improved compared to the early Snap era, but it is still a political and technical choice you should accept consciously.

The question that started this post: if AUR just unpacks .deb files, why am I on Arch?

This is the observation that broke my lazy mental model.

On Manjaro I run roughly 1986 pacman packages, plus a small number of Snaps, plus AUR when needed. In practice, many of the apps I care about are not “native Arch builds.” They are:

• upstream .deb repackaged in the AUR,
• AppImages or GitHub release tarballs,
• or wrapper scripts that download vendor binaries.

Examples from my own experience:

• Google Chrome — official vendor packaging is .deb; AUR often wraps that.
• Cursor — the AUR package I tried wrapped the .deb package. The extracted .deb did not behave as expected on my system, so I moved to the official AppImage instead and wrote about that setup (How to Install Cursor IDE on Manjaro Linux). That experience was a good reminder that “available in AUR” does not always mean “better integrated.”
• Dropbox, Mattermost, VirtualBox, FileZilla — Ubuntu/Debian ecosystems are first-class targets for vendors.

So the honest sentence is not “AUR is useless.” It is:

For mainstream commercial desktop software, Arch-based distros often act as a compatibility layer on top of vendor binaries built for Debian/Ubuntu.

That does not make Manjaro wrong. It means the reason I reach for AUR is narrower than I used to tell myself. I still value AUR for niche tools, patched packages, and community-maintained software that never ships a vendor installer. I just should not pretend it replaces Debian packaging for everything I run daily.

On Kubuntu, the mainstream path is usually:

1

sudo apt install <package>

or a vendor .deb, or snap install, or Flatpak — sometimes two or three options for the same app, which is its own kind of clutter, but at least the vendor tested against that ecosystem.

My real app stack (and what I would verify before switching)

These are the apps I actually care about on a work machine:

Nothing in that table is a guaranteed win for either side without your hardware and your workflow. It is a checklist, not a verdict.

Package management: the trade-off in plain language

Manjaro (rolling, pacman + AUR)

The good parts:

• One primary system package manager with clear dependency semantics.
• Fresh kernels, Mesa, Plasma, and toolchain updates without waiting for an LTS cadence.
• AUR remains excellent for niche software and custom builds.
• mhwd is genuinely helpful for GPU profile selection on install.
• The Manjaro community does something I genuinely appreciate: update announcements in the forum are usually clear, visible, and practical. Before a big stable update, you can read what changed, what broke for some users, and what workaround exists. I have not felt the same sense of one obvious community place around Ubuntu desktop updates.

The rough parts:

• Rolling release means occasional breakage — sometimes in core packages, sometimes in AUR packages after a big update.
• AUR quality varies by maintainer; -git packages and delayed rebuilds happen.
• Vendor apps may still be repackaged deb/AppImage workflows with extra fragility.

Kubuntu (LTS base, apt + snaps + flatpaks)

The good parts:

• Predictable security maintenance for years on an LTS.
• Vendors test against Ubuntu first for many commercial desktop apps.
• Less day-to-day “distro maintenance” if you stay close to the archives.
• Aligns mentally with Ubuntu Server environments many of us use at work (apt, systemd units, paths, docs).

The rough parts:

• Newer software between LTS releases comes from backports, Snaps, Flatpaks, PPAs, or third-party repos — the fragmentation did not disappear; it moved shape.
• Snap policies and Firefox-as-Snap still annoy people for good reasons.
• You can end up with three packaging stories on one machine if you are not disciplined.

KDE experience: finally, not the main argument

For a long time, “I use KDE” was almost synonymous with “I should not use Ubuntu.”

That is no longer the headline.

On my hardware, Manjaro and Kubuntu 26.04 feel like the same Plasma generation: Wayland-first, Qt 6, similar apps, similar settings modules, similar annoyances (themes after upgrades, fractional scaling edge cases, per-app Wayland quirks).

If you are choosing between them only because you want Plasma 6.6 and a modern Wayland session, you are probably choosing for the wrong reason in 2026.

Choose for update philosophy, vendor packaging, GPU behavior on your laptop, and how much time you want to spend being your own distro engineer.

Hybrid graphics: Intel Arc + NVIDIA on a laptop (the part that matters)

This is the section most “distro comparison” posts hand-wave. It matters on my machine.

What I run today (Manjaro)

• Intel Arc integrated GPU + NVIDIA RTX 3050 discrete GPU.
• Manjaro’s mhwd profiles for Intel/NVIDIA hybrid setups (for example video-hybrid-intel-nvidia-prime) install the proprietary NVIDIA stack and configure PRIME-style offload.
• For per-app discrete GPU use on Arch-based systems, prime-run (from nvidia-prime) is the common pattern, alongside environment variables documented on the Arch Wiki PRIME page.

Manjaro has been good enough here for me, not flawless. Hybrid laptops are inherently messy: external displays wired to the dGPU, suspend/resume, Wayland vs X11 behavior, and driver/kernel pairing all still produce forum threads.

What Ubuntu/Kubuntu offers

Ubuntu’s supported path is nvidia-driver-* via ubuntu-drivers, plus nvidia-prime and the prime-select tool:

1
2
3
4

prime-select query
sudo prime-select intel
sudo prime-select nvidia
sudo prime-select on-demand

• intel — power-saving, iGPU drives the session (logout/reboot typically required to switch profiles).
• nvidia — discrete GPU drives the session (higher power draw).
• on-demand — both GPUs available; most apps stay on Intel, and you offload heavy apps to NVIDIA with PRIME Render Offload environment variables (for example __NV_PRIME_RENDER_OFFLOAD=1 and __GLX_VENDOR_LIBRARY_NAME=nvidia), or desktop-specific “run on discrete GPU” integrations where provided.

Wayland + proprietary NVIDIA has improved a lot, but it is still hardware- and driver-version-dependent. On Ubuntu-family systems you may need nvidia-drm.modeset=1 and a recent enough driver for a smooth Plasma Wayland experience. This is an area where VM testing is misleading: my VirtualBox guest uses software rendering (llvmpipe), so it tells me nothing about real PRIME behavior on the laptop.

Is Ubuntu better than Manjaro for hybrid graphics?

The fair answer: neither wins by reputation alone.

• Ubuntu benefits from vendor QA on popular laptop chipsets and enormous documentation surface area.
• Manjaro benefits from newer driver/kernel combos landing sooner — which can fix hardware faster or break it faster, depending on the week.

I would not switch distros based on a blog post claiming a knockout victory. I would switch only after bare-metal tests on my exact machine: external monitor on HDMI/DisplayPort, suspend, on-demand gaming via Lutris, CUDA/ML if needed, and Plasma Wayland session stability.

Rolling release vs LTS: the trade-off I am actually weighing

Manjaro is not “unstable Arch” in meme form, but it is rolling. Updates can require attention: NVIDIA module rebuilds, AUR package rebuilds, occasional Plasma/KWin regressions.

Kubuntu 26.04 LTS is the opposite bet: accept a slower-moving base in exchange for years of security maintenance with less surprise. For a machine that is primarily a work tool — Angular/TypeScript/Nx, browsers, IDEs, calls, family life around it — that stability premium matters more to me now than it did ten years ago.

That is a life-stage preference, not a universal truth. If you enjoy living on the rolling edge and debugging is part of the hobby, Manjaro still shines.

“But I use Ubuntu Server at work”

This is a smaller factor than GPU behavior, but it is real.

Most production Linux I touch is Debian/Ubuntu-shaped: apt, systemd, paths under /etc, packaging docs that assume Ubuntu LTS. Manjaro’s pacman world is not hard to context-switch from, but the friction is constant and low-grade:

• different package names,
• different service layout conventions in third-party docs,
• different “just apt install it” copy-paste moments in tutorials.

Matching desktop and server is not required for competence. It is a quality-of-life bonus that accumulates over years.

What I am not claiming

To keep this post useful instead of preachy:

• Manjaro did not “go bad.” It still delivers what I wanted when I switched.
• Kubuntu is not automatically better because it is LTS and popular.
• AUR is not dead — it is just not the universal advantage I once mentally assigned it for vendor apps.
• Snap is not “fixed” for everyone — it is more mature, and still a value judgment.
• I have not migrated yet. VM success is necessary but not sufficient.

What I am doing next

This is my practical plan, and maybe yours if you are in the same boat:

1. Keep Kubuntu 26.04 in a VM for workflow tests (Cursor, Node projects, browsers, Mattermost, basic dev containers).
2. Do not rush the migration. I would rather keep testing in a VM for a while than turn my main work machine into a weekend experiment.
3. Switch when I have enough quiet time to reinstall and configure all my systems properly, not just one laptop in a hurry.
4. Run the checklist that VM cannot answer after the switch: hybrid GPU, battery life, fan curves, external monitor behavior, suspend, Lutris, VPN/proxy tooling, Dropbox sync, and Wayland + NVIDIA on real sessions.

If those pass after the move, switching was rational. If they do not, I will at least know exactly which trade-off was not worth it.

Closing thought

The Linux desktop did change while I was on Manjaro. Ubuntu and Kubuntu in 2026 are not the slow KDE + old kernel experience that pushed me toward rolling release years ago.

But the question was never “Which distro wins in 2026?”

The question is: what problems do I still need a rolling Arch-based distro to solve for me — and what problems would an LTS Ubuntu base solve better today?

Right now, my answer is: maybe fewer of the first than before, and more of the second than I expected — especially around vendor packaging and maintenance cost. The laptop GPU test is still the gate.

I will post a follow-up if I do the bare-metal trial. If you have switched either direction recently on Intel + NVIDIA hybrid hardware, I would genuinely like to hear what broke and what did not.

References and further reading

• Kubuntu 26.04 LTS release notes
• Arch Wiki: PRIME
• My earlier post: How to Upgrade Ubuntu Kernel (Mainline)
• My earlier post: How to Install Cursor IDE on Manjaro Linux