OpenMetal v5 Server Hardware
Intel® Xeon® 6 performance. Transparent pricing. Fully yours.
The next generation of OpenMetal infrastructure is here!
Built on Intel® Xeon® 6 processors, DDR5-6400 memory, and high-endurance NVMe storage. More compute per node, more memory bandwidth, and the predictable, transparent pricing the hyperscalers won’t give you.
What’s New in OpenMetal v5
OpenMetal v5 is a generational platform update, not a cosmetic refresh.
Intel® Xeon® 6 (Granite Rapids)
Higher core counts per socket, stronger per-core performance, and improved power efficiency over prior generations, a more capable foundation for compute-intensive workloads.
DDR5-6400 memory across all SKUs
A significant jump in memory bandwidth and performance consistency, especially for virtualization-heavy and data-intensive environments.
High-endurance NVMe storage
Standardized on Micron 7500 MAX, selected for sustained performance and endurance under real production load: databases, virtualization, and write-intensive services.
Confidential computing, included
Every v5 system ships with TPM support and is built to support Intel® TDX / SGX where memory topology requirements are met. It is capability the hyperscalers typically charge a premium for, here as standard.
The v5 platform also includes dedicated GPU servers for AI, machine learning, and high-performance computing. Same Intel Xeon 6530P foundation, same DDR5-6400 memory, same 40Gbps private networking — with NVIDIA’s latest GPU hardware and no shared hypervisor between your workload and the silicon.
The v5 catalog represents a step change in what we’re able to deliver to customers. Intel’s Xeon 6000 P-series unlocks a new level of per-core performance, and pairing that with DDR5-6400 memory means customers running data-intensive workloads will see real-world gains from day one. We engineered this lineup to handle what’s coming next, not just what customers are running today.
Jamie Tischart, CTO @OpenMetal
Contact us about v5 Hardware
Use the form below to contact us about the newly released v5 hardware and which configurations you are interested in.
Frequently Asked Questions
Is v5 available now?
v5 compute servers are available now in Ashburn, VA (US East) as bare metal or hosted private cloud. The RP6000 (NVIDIA RTX PRO 6000 Blackwell) and H200 GPU servers are also available now in Ashburn. Compute can be pre-ordered for Los Angeles, Amsterdam, and Singapore.
How is v5 priced?
Fixed, transparent monthly pricing with no metered egress surprises. See full pricing in the catalog or talk to an account manager.
Hosted Private Cloud catalog >>
Can I order multiple systems?
Yes. Your account manager will confirm availability and lead times for larger deployments.
Can configurations be customized?
Memory and storage layouts can often be adjusted. Talk to your account manager about specific requirements.
Is v5 compute only for bare metal?
No. v5 compute deploys two ways: as single-tenant bare metal, or as a Hosted Private Cloud of three identical v5 servers hyper-converged on OpenStack and Ceph. Same hardware, your choice of model.
What GPU servers are available on v5?
The v5 catalog includes two GPU server lines, both built on the same dual Intel Xeon 6530P and DDR5-6400 foundation as the compute tier. The RP6000 pairs that platform with one or two NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs (96GB GDDR7 each), suited for inference, fine-tuning, rendering, and mixed AI and visualization workloads. The H200 pairs it with one or two NVIDIA H200 NVL GPUs (141GB HBM3e each) and includes a five-year NVIDIA AI Enterprise software subscription, targeting large-model training and memory-bandwidth-bound inference. Both are available now in Ashburn, VA.
What’s the difference between the RP6000 and H200?
The RP6000 is built for high-VRAM workloads at a favorable cost per GB — inference serving, fine-tuning, rendering, and pipelines that mix AI and visualization. The H200 is built for workloads where memory bandwidth and capacity are the bottleneck: large-model training, large-context inference, and memory-bound HPC. The H200 also includes a five-year NVIDIA AI Enterprise software subscription. Talk to an account manager if you’re unsure which fits your workload.
Can I build a multi-node GPU cluster?
Yes. Multiple GPU servers can be interconnected over OpenMetal’s 40Gbps private network to build multi-node training or inference clusters. The GPU servers support PyTorch, TensorFlow, JAX, and Hugging Face Transformers. Talk to an account manager about cluster configurations and lead times.
Is a proof-of-concept available for GPU servers?
Yes. PoC deployments are available so your team can validate workloads on dedicated hardware before committing. Talk to sales to get started.
Do all v5 compute servers support confidential computing?
All ship with TPM. Intel SGX is enabled across the v5 line. For Intel TDX (VM-level confidential computing), the XL v5 ships TDX-ready at 1 TB and a BIOS toggle activates it, while the Large and Medium v5 reach TDX via a RAM upgrade to 1 TB (full DIMM population). Ask your account manager about the right configuration.