← Back to GPU Leasing Hopper Architecture Most Popular

NVIDIA H100

The global standard for enterprise AI training and inference.

Leasing Price - All-inclusive $2.75 / GPU / hour Includes power, cooling, infrastructure and 24/7 management
Request a Leasing Quote Contact Sales
Overview

NVIDIA H100 SXM5 / PCIe 5.0

The NVIDIA H100 is the most widely deployed enterprise GPU for AI. Hopper architecture with 4th-generation Tensor Cores and Transformer Engine delivers massive performance for LLM training and inference.

Technical Specifications

Leasing Price
$2.75 / GPU / hour
GPU Architecture
NVIDIA Hopper (GH100)
CUDA Cores
16,896
Tensor Cores
4th Gen 528 cores
GPU Memory
80 GB HBM3
Memory Bandwidth
3.35 TB/s
FP16 Performance
1,979 TFLOPS
FP8 Performance
3,958 TFLOPS
NVLink
4th Gen 900 GB/s
TDP (SXM5)
700 W
TDP (PCIe)
350 W
MIG Support
Up to 7 instances
Use Cases

What is the NVIDIA H100 used for?

Large language model training
LLM inference at scale
Generative AI
Scientific computing
Drug discovery
Recommendation systems
Is this GPU right for you?

Why lease the NVIDIA H100?

The H100 is the safest and most supported choice for serious AI infrastructure. Excellent software ecosystem proven at scale globally.

Not sure? Contact our team and we will help match your workload to the right GPU.

[email protected]
Available Lease Terms
1 YearFull upfront payment
2 YearsFull upfront payment
3 YearsFull upfront payment
4 YearsFull upfront payment
5 YearsFull upfront payment
6 YearsFull upfront payment
7 YearsFull upfront payment
Server Configuration

What is Inside Each NVIDIA H100 Server

Purpose-built for NVIDIA H100 compute. Enterprise-grade throughout, managed 24/7. You lease complete servers, not shared resources.

GPU - NVIDIA H100 x 8

Hopper Architecture

GPUs per Server
8 x NVIDIA H100
Memory per GPU
80 GB HBM3
Total GPU Memory
640 GB
Bandwidth per GPU
3.35 TB/s
Interconnect
4th Gen 900 GB/s
TDP per GPU
700 W
Total Server TDP
5,600 W total

System - CPU, Memory and Infrastructure

4U Rackmount

CPU
2x AMD EPYC 9965 (192-core)
System Memory
2 TB DDR5 6400 ECC
Memory Config
64x 32 GB Registered ECC
Storage
2x 15.36 TB U.2 NVMe PCIe 4.0
Network
2x 200 GbE InfiniBand HDR
Power Supply
10,000W (5x 2,000W Titanium)
Management
IPMI - 24/7 Remote Hands
Form Factor
4U Rackmount
🔒

Exclusively yours. Every server is dedicated to a single customer - no shared tenancy. Archean Core AS owns and manages all hardware. You get full control throughout your lease.

Pricing Calculator

Estimate Your Leasing Cost

$2.75 USD per GPU per hour. Lease by the server - 1 server = 8 GPUs. 2 years: 10% off - 3+ years: 15% off - 7 years: 30% off. Full upfront payment required.

iNVIDIA H100 servers contain 8 GPUs each. You lease 1 complete server at a time. Each server gives you dedicated access to 8 x NVIDIA H100 GPUs.
1 server = 8 GPUs - Max 125 servers (1,000 GPUs)

Estimated Total Cost

-

Price per GPU$2.75 / hour
Effective price$2.75
1 server - 8 GPUs - 1 year70,080 hrs
Per server / month-
Per server / year-

Prices in USD. Final pricing confirmed in your contract. Full upfront payment required.

Place Order Below
Place Your Order

Order NVIDIA H100 Leasing

Fill in your details and our team will send a formal contract and invoice to your email - typically within one business day.

Full upfront payment required for all lease terms. A formal invoice will be sent with your contract before any payment is due.
✓

Order Request Received

Our team will prepare a formal contract for your NVIDIA H100 lease and send it to your email within one business day.

[email protected]

Order Summary

GPUNVIDIA H100
Price per GPU$2.75 / hr
GPUs per server8
Servers-
Total GPUs-
Lease term-
Total hours-
Estimated Total -
✓Contract within 1 business day
✓Formal invoice included
✓No payment until contract signed
✓Dedicated hardware - no shared tenancy
✓99.9% uptime SLA included