Local AI rig: 32 GB GPU build
32 GB on one card gives 32B-class models room for long conversations and runs some larger MoE models with offload. See exactly what fits on our 32 GB tier page.
We don't show prices here because we only print prices we checked the same day, and Amazon prices change often. Click through to see today's price.
Parts list
| Part | Pick | Why | Link |
|---|---|---|---|
| GPU | GeForce RTX 5090 (32 GB) | 32 GB GDDR7 on a 512-bit bus, 575 W total graphics power. NVIDIA lists a 1000 W power supply (NVIDIA). | Check price on Amazon |
| GPU (AMD option) | Radeon AI PRO R9700 (32 GB) | 32 GB GDDR6, 256-bit, 300 W total board power (AMD). Check your app's AMD support first. | Check price on Amazon |
| System RAM | 64 to 128 GB DDR5 | Needed if you offload big MoE models' experts to the CPU. See MoE expert offload. | Check price on Amazon |
| Storage | 2 TB NVMe SSD | Room for several large model files. | Check price on Amazon |
| Power supply | 1000 W or larger ATX 3.1 PSU | Matches NVIDIA's listed 1000 W for the RTX 5090. | Check price on Amazon |
| Case | Full or mid tower that fits the card's length | High-end cards are long and thick; check the card's dimensions against the case. | Check price on Amazon |
Amazon links are affiliate links: we may earn a commission at no cost to you. Disclosure.
Next steps
- Compare 32 GB cards in our 32 GB buying guide and see the RTX 5090 page.
- Check your exact model in the VRAM calculator.
- Other builds: 16 GB budget rig · 24 GB rig.
- Only need it a few hours a month? Renting a GPU may be cheaper.