OpenMetal GPU Servers are getting a major upgrade as the company expands its v5 hardware catalog with new NVIDIA-powered systems built for AI, machine learning, and high-performance computing.
The company has introduced two private GPU server lines: RP6000 and H200. The RP6000 is based on NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, while the H200 uses NVIDIA H200 NVL GPUs. Both are designed to give customers dedicated GPU performance, single-tenant infrastructure, and predictable monthly billing.
The launch strengthens OpenMetal’s v5 platform, which already includes Intel Xeon 6000-series processors, DDR5-6400 memory, PCIe 5.0 connectivity, NVMe storage options, and 40Gbps private networking.
OpenMetal GPU Servers Target AI and HPC Teams
OpenMetal GPU Servers are aimed at organizations that need high-performance computing power without relying on shared cloud infrastructure.
Each system runs on dual Intel Xeon 6530P processors and includes 1TB of DDR5-6400 memory as standard. Customers can expand memory to 2TB, giving teams more room for demanding AI and technical workloads.
The servers are built for customers that want direct control over their hardware, clearer pricing, and consistent performance for serious workloads.
OpenMetal GPU Servers Add RTX PRO 6000 Blackwell
The RP6000 line supports one or two NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs per server.
Each GPU offers 96GB of GDDR7 memory, 1.79 TB/s of memory bandwidth, and 24,064 CUDA cores. This makes the RP6000 useful for AI inference, model fine-tuning, rendering, visualization, and mixed AI production pipelines.
The large VRAM capacity gives teams strong performance for workloads that need powerful GPUs but also require cost control.
OpenMetal GPU Servers Bring H200 for Large Models
The H200 server line is built for heavier AI workloads.
Each NVIDIA H200 NVL PCIe GPU includes 141GB of HBM3e memory and 4.8 TB/s of memory bandwidth. This makes the system suitable for large-model training and memory-heavy inference tasks.
Every H200 server also includes a five-year NVIDIA AI Enterprise software subscription, giving teams access to a supported AI software stack for production deployments.
OpenMetal GPU Servers Offer Dedicated Bare Metal Control
OpenMetal GPU Servers are delivered as single-tenant bare metal infrastructure.
That means customers do not share a hypervisor with other users. This setup can improve performance consistency, security, and operational control for teams running sensitive or resource-heavy workloads.
The systems support popular AI and machine learning frameworks, including PyTorch, TensorFlow, JAX, and Hugging Face Transformers. Customers can use the servers independently or connect them into multi-node GPU clusters through OpenMetal’s private network.
Why OpenMetal GPU Servers Matter
OpenMetal GPU Servers arrive as demand for private AI infrastructure continues to grow.
Many businesses want GPU acceleration but also want to avoid unpredictable usage-based costs and shared infrastructure limits. OpenMetal is positioning its new RP6000 and H200 servers as an option for teams that need dedicated resources and transparent pricing.
The new servers are available now in OpenMetal’s US East data center in Ashburn, Virginia. Additional locations are planned, and proof-of-concept deployments are available for teams that want to test workloads before making a longer commitment.
As AI and HPC workloads become more demanding, OpenMetal GPU Servers give enterprises, SaaS firms, hosting providers, and managed service providers another way to build powerful private infrastructure with greater control.
