I want to make powerful local AI easier for everyone to use.

I'm raising money for an NVIDIA RTX 5090—or a comparable high-end GPU—to help me continue an open-source, educational project around local AI.

This started as a personal, off-hours project.

I wanted to understand what it actually takes to run modern AI models yourself: on a Mac, on Windows, on inexpensive hardware, and eventually on increasingly capable local machines.

The more I learned, the more I realized that the technology is becoming incredibly powerful, but getting started can still be unnecessarily complicated.

So I'm trying to make it simpler.

What I've been building

I've been writing practical articles and publishing open-source repositories that explain how to run LLM servers locally across different platforms.

Some of the projects I've already built include:

The idea behind all of them is the same:

Take something complicated, figure it out, simplify it, and share what I learn.

I don't have all the answers. I'm learning as I go, and that's intentional.

AI is changing extremely quickly. I think one of the best ways to understand where it's going is to actually build things with it, run the models yourself, measure what works, and share the results.

Why I need a 5090

I'd like to take the project much further.

A high-end NVIDIA GPU would give me enough local compute to experiment with substantially larger and more capable models and to explore workloads that simply aren't practical on the hardware I currently have access to.

I want to explore the full local AI stack:

text → vision → audio → image generation → video generation → agents → APIs → self-hosted applications

And then turn what I learn into articles, examples, and open-source repositories that other people can use.

There is something powerful about being able to download a project, run it on your own machine, and understand exactly what is happening.

I want to help make that experience normal.

Why crowdfunding?

The honest answer is that I can't afford a GPU like the RTX 5090 myself.

This is an off-hours project. I also have a 2.5-year-old son, so buying several thousand euros of computer hardware for a personal project isn't something I can reasonably prioritize.

At the same time, I don't want that to be the reason the project stops progressing.

So I'm asking whether the people who find this work interesting—or useful—would like to help fund the hardware.

This isn't a company fundraising round.

There isn't a business behind this.

It's one person spending his spare time learning, building, writing, and sharing.

And I'm hoping a community of people who are excited about local AI might want to help me keep going.

What happens if we reach the goal?

I'll buy the best suitable GPU within the campaign budget, ideally an RTX 5090 or comparable hardware.

Then I'll put it to work.

I'll continue publishing what I learn and building projects around local AI, with an emphasis on making them understandable and practical rather than overly complicated.

I'll also share the journey.

What works.

What doesn't.

How fast different models run.

What hardware actually matters.

How to deploy these systems.

And, most importantly, what ordinary people can realistically do with AI running on hardware they control themselves.

You don't have to donate.

If you like the idea, that's enough.

If you've learned something from one of my articles or repositories, a donation is one way to help me keep doing this.

A small contribution is useful.

A large contribution is obviously very useful.

A share is useful too.

And if you happen to have a suitable GPU that you're no longer using, I'd be very happy to consider a hardware donation as well.

Why I'm doing this

I think local AI is going to become increasingly interesting.

Not because everyone needs to own the fastest GPU.

But because people should have the option to understand, experiment with, and run powerful AI systems themselves.

The hardware will keep getting better.

The models will keep getting better.

The tooling will keep getting better.

I want to help make the path from "I've heard about this" to "I'm running it myself" as short and approachable as possible.

That's what I'm working toward.

If you'd like to help me get the hardware to take the next step, thank you.

I'll do my best to turn that support into something useful that others can learn from and build on.

And I'll keep sharing what I learn along the way.

GitHub: https://github.com/kibotu

Article: Two Paths to Local LLM Servers — Windows/NVIDIA vs. Mac/Apple Silicon

Thank you for being part of the journey.