The guide explains three ways to install NVIDIA’s proprietary driver on Ubuntu 22.04 LTS—using the “Additional Drivers” GUI, installing the recommended package via the terminal with ubuntu‑drivers/apt, or running NVIDIA’s own .run installer when a newer version is required. It highlights common obstacles such as Secure Boot blocking unsigned modules, kernel updates breaking manually installed drivers, and the need for matching kernel headers and DKMS support. After any method, you verify success by running nvidia‑smi, which should list your GPU and driver version. Finally, it advises disabling Secure Boot or enrolling a key, using Ubuntu’s repository drivers for stability, and installing nvidia‑prime on hybrid‑graphics laptops to avoid future issues.

How to Install NVIDIA Drivers on Ubuntu 22.04 LTS

You’ll learn three ways to get the proprietary NVIDIA graphics driver working on Jammy Jellyfish, see why each method matters, and avoid the common “driver vanished after a kernel upgrade” nightmare. I’ll point out the tools that actually help and call out the ones that just waste your time.

Use Ubuntu’s “Additional Drivers” GUI (the lazy but reliable route)

1. Open Software & Updates from the Activities menu, switch to the Additional Drivers tab.
   – The system scans your hardware, lists every driver it can install, and marks the recommended one.
2. Select the entry that says “NVIDIA driver metapackage … (proprietary, tested)”.
3. Click Apply Changes and wait for apt to pull the packages.
4. Reboot so the new kernel module loads.

Why this works: Ubuntu already bundles the correct version of the driver in its repositories, and it automatically pulls matching kernel headers. I’ve seen people manually download a .run file only to end up with a black screen after the next update – the GUI method sidesteps that mess.

Pro tip: If the list is empty, you probably have Secure Boot enabled. Ubuntu refuses to load unsigned NVIDIA modules under Secure Boot, so either disable it in your BIOS or enroll a Machine‑Owner Key (MOCK) – the latter is overkill for most home users.

Screenshot: Additional Drivers tab showing NVIDIA driver selection

Install from the command line with apt (the “I like terminals” approach)

Open a terminal and run:

sudo ubuntu-drivers devices

The command prints something like:

== /sys/devices/pci0000:00/0000:00:1f.0 ==
modalias : pci:v000010DEd00001E04sv00001458sd000037A5bc03sc00i00
vendor : NVIDIA Corporation
model : GA106 [GeForce RTX 3060]
driver : nvidia-driver-525 - distro non-free recommended

Take note of the recommended package name (here nvidia-driver-525). Then install it:

sudo apt update
sudo apt install nvidia-driver-525

Why you should do this instead of blindly installing the latest version: Ubuntu’s repository only ships drivers that have been tested against the current kernel series. Installing a newer driver from the “graphics‑drivers” PPA can break things after a routine kernel upgrade.

After installation, run:

sudo systemctl reboot

When you log back in, nvidia-smi should show your GPU and driver version.

Real‑world note: I once upgraded to kernel 6.2 on Jammy, kept the driver from a PPA, and the X server refused to start. Rolling back to the repository driver fixed it in minutes.

Manual .run installer from NVIDIA (the “I trust the vendor more than Ubuntu” method)

Warning: This is rarely needed and often leads to headaches. Use only if you need a brand‑new driver that Ubuntu hasn’t packaged yet.

1. Download the latest Linux 64‑bit .run file from https://www.nvidia.com/Download/index.aspx.
2. Switch to a text console: Ctrl+Alt+F3, log in, and stop the display manager:

sudo systemctl isolate multi-user.target

Stopping the X server prevents the installer from trying to replace an active module.

3. Make sure you have the kernel headers that match your running kernel:

sudo apt install build-essential linux-headers-$(uname -r)

4. Run the installer:

sudo sh NVIDIA-Linux-x86_64-*.run

Answer “yes” to the DKMS integration prompt – it will rebuild the module automatically after each kernel update.

5. Re‑enable the graphical target and reboot:

sudo systemctl start graphical.target
sudo reboot

Why you might still end up with trouble: The installer overwrites Ubuntu’s nvidia-dkms package, so future apt upgrades may try to reinstall the older repository driver and clash with your manual files. You’ll spend more time fighting package conflicts than actually gaming.

Verify the driver is active

Regardless of how you installed it, run:

nvidia-smi

You should see a table listing your GPU, driver version, and CUDA cores. If you get “Command ‘nvidia‑smi’ not found”, the driver didn’t load – double‑check Secure Boot or that you rebooted after installation.

Common pitfalls and how to avoid them

• Secure Boot – unsigned modules are blocked. Either turn it off or enroll a key (the former is quicker).
• Kernel updates – If you used the .run installer, remember to rerun it after each kernel change unless DKMS handled it. The apt method does this automatically.
• Hybrid graphics laptops – On systems with both Intel and NVIDIA GPUs, install nvidia-prime (sudo apt install nvidia-prime) and switch between modes with the “Prime Profiles” app.

That’s it. Pick the route that matches your comfort level, verify with nvidia-smi, and you should be back to smooth 4K gaming or CUDA crunching in no time.