consumer gpu

NVIDIA GeForce RTX 5090

The on-ramp: a single desktop card for prototyping with open models

$2,399

Street price as of mid-2026; $1,999 MSRP at Jan 2025 launch, elevated since by sustained demand.

Specs

GPUs in this unit1
Total GPU memory32 GB
Memory bandwidth (per GPU)1,792 GB/s
Total power draw575 W
Compute (dense BF16/FP16)210 TFLOPS

GDDR7 memory, 21,760 CUDA cores, single PCIe card, standard desktop power connectors.

What this actually means for you

This is NVIDIA's fastest gaming/creator card, repurposed for AI. It's the cheapest way to get real hands-on experience running an open model locally, but it's a single consumer card with no ECC memory and no multi-GPU networking — fine for prototyping, not for serving paying customers.

Solo capacity: at the standard FP16/BF16 rule of thumb (2 GB of GPU memory per billion parameters, plus 20% overhead for context), this unit's 32 GB can hold an open-source model up to roughly 13 billion parameters before you'd need more hardware.

Power & running cost

Continuous draw

0.57 kW

Monthly electricity

$78

Everyday comparison

~48% of one home

Run 24/7, this unit uses 13.8 kWh/day — about the same continuous power draw as 0.5x an average U.S. home (~29 kWh/day, EIA). Monthly cost assumes the 2026 U.S. residential average of 18.8¢/kWh.

Request this build

No payment, no account — just tell us your name and we'll follow up with a real quote.