Despite my better judgement, I just bought an RTX 5090 for about $3,200 with tax. This is by far the most expensive graphics card I’ve purchased since I’ve always stayed away from buying high end cards. You get heavy diminishing returns on price on the high end of hardware, but I thought it would be worth it this time. It also seemed like a good time to buy since RAM prices were spiking due to AI data centers scooping up the supply, but the price of GPUs hadn’t really increased yet.

My main reason for getting it was to experiment with AI models locally. I’ve been regularly using Copilot in agent mode for programming, and various other tools for image and chat, but I find local models appealing. I don’t particularly want to tell Sam Altman the details of some medical condition, or other private information. So using a local LLM for those questions makes sense and I’m curious to see how much I could push local models with something this beefy. I was running an RTX 2070 Super with 12GB of VRAM that I bought in 2019 and had been tinkering with small models on it. The results were mediocre at best, but I bet fitting much larger models on this card will remedy that.

There’s also the gaming aspect of it. Again, the 2070 Super is fine. I’ve never been obsessed with graphics, and I play a lot of old games anyway. The most demanding game I play regularly is Cyberpunk 2077 and it looks fine on the old card. After firing up the 5090 I cranked the ray tracing in Cyberpunk to max and…yes the reflections and lighting were a bit better. Not enough to justify the hardware upgrade certainly, so it was a good thing that wasn’t my reason for buying it.

So this card gave me a lot of room to experiment with local models, while futureproofing on the gaming side of things. The 2070 Super worked fine for 6 years, so I’ll probably get a good 6 years out of this one too.

Update March 2026: They’re going for $1000 more now. Absolutely called it on the price spike and bought at the perfect time.