Metal Cloud
Dedicated GPU Bare Metal for Enterprise AI
Access dedicated NVIDIA H100 & H200 bare metal servers with full infrastructure control
Purpose-Built Infrastructure for Large-Scale AI Workloads
NVIDIA H100 & H200 Bare Metal
Dedicated GPUs with full performance, ideal for large model training and inference
Ultra-Fast Deployment
Rapid GPU server provisioning within days, reducing deployment time
High-Bandwidth Networking
400Gbps InfiniBand / RoCE-ready architecture optimized for distributed training
Scalable Multi-Node Architecture
Multi-node architecture with managed Kubernetes and
high-performance shared storage
Reserve your cluster from today
Enterprise-grade NVIDIA-designed GPU clusters with dedicated resources
GPU VM
Specification
Price
1x
GPU B300
288GB GPU Memory
192 GB RAM | 32 cores CPU | 100 GB Block Storage
6th Gen Intel Xeon Scalable Processors
Pre-Order Available
2x
GPU B300
576GB GPU Memory
384 GB RAM | 64 cores CPU | 200 GB Block Storage
6th Gen Intel Xeon Scalable Processors
Pre-Order Available
4x
GPU B300
1152 GB GPU Memory
768 GB RAM | 128 cores CPU | 400 GB Block Storage
6th Gen Intel Xeon Scalable Processors
Pre-Order Available
8x
GPU B300
2.2 TB GPU Memory
1536 GB RAM | 256 cores CPU | 800 GB Block Storage
6th Gen Intel Xeon Scalable Processors
Pre-Order Available
Why Choose Metal Cloud?
Scalable infrastructure designed to power next-generation AI workloads
Dedicated GPU Resources
Your workload runs on fully dedicated GPU memory and compute power
Complete System Control
Full root access to your server, with complete control over the operating system, drivers, and configurations
High-Speed Cluster Networking
Built for large-scale training, the infrastructure delivers ultra-fast communication between GPU nodes
Shared Storage
All GPU nodes access a single shared dataset in real time, with no duplication, no conflicts, and no lost checkpoints
Use Cases
Designed for High-Performance and AI-Driven Workloads
Build and Train Large AI Models Faster
Get the infrastructure you need to train and fine-tune large language models efficiently
Serve AI Models at Scale with Low Latency
Deploy dedicated inference nodes designed for real-time, high-throughput, low-latency performance
Explore and Experiment Without Limits
Take full control of your GPU stack all without queues or platform restrictions