moonsun

Pricing

Our Dedicated Compute & On Demand GPU Instances are built to give AI teams access to the newest generation of NVIDIA GPU's in regional edge sites to streamline your performance across multiple workloads.

Dedicated
card icon

1/2 Megawatt

288 GPU's

Dedicated NVIDIA Blackwell GB 300 GPU's

Key features:
288 NVIDIA GB300 Blackwell Ultra GPUs
800 Gbps Spectrum or Infiniband RDMA
Bare Metal Access with full OS control
Liquid to chip cooling
Zero Data Egress Fees
Reserve Now

Reserve Capacity Now. No Obligation

card icon

1 Megawatt

576 GPU's

Dedicated NVIDIA Blackwell GB 300 GPU's

Key features:
576 NVIDIA GB300 Blackwell Ultra GPUs
NVIDIA NVL72 Rack-Scale Architecture
Non-Blocking Fat-Tree Topology
135kW+ High-Density Power
Urban Edge Proximity <1 ms latencies
Reserve Now

Reserve Capacity Now. No Obligation

Popular
card icon

2 Megawatt

1,152 GPU's

Dedicated NVIDIA Blackwell GB 300 GPU's

Key features:
1152 NVIDIA GB300 Blackwell Ultra GPUs
Sovereign Private Deployment
Redundant Liquid-Cooled Infrastructure
Bespoke Hardware Tuning
Unrestricted Bare Metal & Zero Egress
Reserve Now

Reserve Capacity Now. No Obligation

On-Demand
card icon

Node

8 GPU's

Dedicated NVIDIA Blackwell GB 300 GPU's

Key features:
8 NVIDIA GB300 Blackwell Ultra GPUs
Billed per hour with zero lock-in
PRe-configured AI/ML framework containers
Instant Provisioning, scalable NVMe storage
Zero Data Egress Fees
Deploy Now

Launch Instance. Pay as you go.

card icon

Rack

72 GPU's

Dedicated NVIDIA Blackwell GB 300 GPU's

Key features:
72 NVIDIA GB300 Blackwell Ultra GPUs
NVIDIA NVL72 Rack-Scale Architecture
Fractional or short-term monthly commitments
Bare Metal or Managed Kubernetes environments
Urban Edge Proximity <1 ms latencies
Deploy Now

Launch Instance. Pay as you go.

Popular
card icon

Pod

288 GPU's

Dedicated NVIDIA Blackwell GB 300 GPU's

Key features:
288 NVIDIA GB300 Blackwell Ultra GPUs
800 Gbps Spectrum or Infiniband RDMA
Dynamic workload scaling for burst capacity
High-Performance Parallel File System
Unrestricted Bare Metal & Zero Egress
Deploy Now

Launch Instance. Pay as you go.

Prices exclude any applicable taxes.

Created by potrace 1.16, written by Peter Selinger 2001-2019
Created by potrace 1.16, written by Peter Selinger 2001-2019

Frequenlty asked questions:

Are there really zero data egress fees?

+

How quickly can I get access to On-Demand versus Dedicated instances?

+

What networking infrastructure supports the NVIDIA GB300 Blackwell GPUs?

+

Can I seamlessly migrate my workloads from On-Demand to a Dedicated cluster as my project scales?

+