datacenter gpu

NVIDIA H200 SXM

Same compute as the H100, with far more memory to work with

$35,000

NVIDIA does not publish a list price; this reflects typical broker/OEM pricing in 2026.

Specs

GPUs in this unit1
Total GPU memory141 GB
Memory bandwidth (per GPU)4,800 GB/s
Total power draw700 W
Compute (dense BF16/FP16)989 TFLOPS

HBM3e memory, SXM5 socket, drop-in replacement for H100 in the same server chassis.

What this actually means for you

Identical compute speed to the H100, but with almost twice the memory and 43% more memory bandwidth (how fast data moves in and out of memory — the thing that actually limits how fast a big model responds). This is the default building block behind most of the multi-GPU builds below.

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 141 GB can hold an open-source model up to roughly 59 billion parameters before you'd need more hardware.

Power & running cost

Continuous draw

0.70 kW

Monthly electricity

$95

Everyday comparison

~58% of one home

Run 24/7, this unit uses 16.8 kWh/day — about the same continuous power draw as 0.6x 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.