Thinking about starting a homelab? Follow one user’s journey from a simple Raspberry Pi rack to building a custom router and exploring uses for a powerful GPU.

I spent years in the IT world, starting way back in the 90s. But about a decade ago, I traded my command line for a cap and gown, moving into university administration. I’m a Dean now, but I never quite managed to shake the tech bug. It’s a part of who I am.

So, when Amazon Prime Day rolled around, I saw a deal on a little 4U server rack and thought, “Why not?” It seemed like the perfect home for my growing collection of Raspberry Pis.

That was the start. It was a small, simple project. But if you’ve ever tinkered with tech, you know how this goes. One small project is never just one small project.

From a Small Rack to a Bigger Plan

That little 4U rack filled up faster than I expected. Soon, I was browsing eBay for micro PCs and upgrading to a bigger 8U rack. As the hardware grew, so did the complexity. I realized I needed a way to keep track of it all—what was connected to what, which IP addresses were assigned where, and what my future plans were.

This led me down a completely different rabbit hole: Obsidian. It’s a note-taking app, but it’s so much more than that. I started using it to document my entire homelab. I created diagrams, network maps, and notes on every piece of hardware and software. It’s my single source of truth. If you’re starting a homelab, my best advice is to document your setup from day one. You’ll thank yourself later.

My latest project? Ditching my ISP’s router. I have a 2Gbps internet plan from Xfinity, but the mesh router I was using just couldn’t keep up. It was bottlenecking my speeds, which is the kind of problem a good homelab is built to solve. So, I’ve got a new, dedicated machine on the way to build my own router using OPNsense, a powerful open-source firewall.

The Big Question: What Do I Do With This GPU?

In the middle of all this planning, I have one piece of hardware that feels like a beautiful, unsolved puzzle. My main server, a Dell Precision running Proxmox, is hosting my TrueNAS storage setup. And inside that machine is a beastly NVDIA RTX A5000 GPU.

It’s powerful. Maybe a little too powerful for what I’m doing right now, which is… nothing. And that feels like a waste.

So, what can you actually do with a high-end GPU in a home server? I’ve been digging into this, and it turns out, there are some pretty cool answers.

• Media Server Transcoding: This is the most common use case. If you run a media server like Plex or Jellyfin, a GPU can handle video transcoding on the fly. That means if you’re streaming a 4K movie to your phone, the GPU can convert the file to a smaller format without breaking a sweat, leaving your main CPU free for other tasks.

• Running AI Models: This is where things get really interesting. You can use a powerful GPU to run your own local AI and machine learning models. Think running a private version of a large language model (like ChatGPT) or setting up Stable Diffusion to generate AI images right from your own server. It’s a great way to experiment with AI without relying on cloud services.

• GPU Passthrough for Virtual Machines: With Proxmox, I can dedicate the entire GPU to a specific virtual machine (VM). This would let me spin up a high-performance Windows or Linux VM and use it for things that need serious graphics power, like video editing, 3D rendering with software like Blender, or even cloud gaming. I could stream games from my server to any screen in the house.

• Data Science and Computation: While I’m not a data scientist, a GPU like the A5000 is built for heavy-duty computation. It could be used for scientific research, complex simulations, or massive data analysis projects.

For now, I’m leaning toward setting it up for Plex transcoding and starting to play around with some local AI models. It feels like the right mix of practical and experimental—which is what a homelab is all about. It’s a personal space to learn, build, and solve your own unique challenges. It all starts with one small project, and from there, the journey is up to you.