Supermicro ARS-111GL-NHR2 1U H100 NVL PCIe GPU Server (Rev 2)

The Supermicro ARS-111GL-NHR2 is a 1U rack-mounted GPU server engineered specifically to support the NVIDIA H100 NVL PCIe GPU in enterprise datacenter and AI infrastructure environments. Built on AMD EPYC processor architecture, this second-revision platform is optimized for high-throughput AI inference and training workloads that demand both computational density and PCIe Gen 5 bandwidth. The system integrates Supermicro's advanced thermal management design to sustain full GPU and CPU performance within a compact 1U chassis footprint, making it well-suited for organizations seeking to maximize rack density without sacrificing sustained compute throughput.

The H100 NVL PCIe form factor supported by this server delivers NVIDIA's Hopper GPU architecture, featuring Transformer Engine capabilities and fourth-generation Tensor Cores designed to accelerate large language model inference, generative AI, and scientific simulation at scale. The ARS-111GL-NHR2 leverages PCIe connectivity rather than NVLink fabric, enabling broad compatibility with standard datacenter networking and storage infrastructure while still delivering exceptional per-GPU performance for enterprise AI pipelines. The platform supports substantial system memory capacity alongside NVMe storage options, ensuring that data ingestion and model serving pipelines are not bottlenecked at the host level.

As a Rev 2 iteration, the ARS-111GL-NHR2 incorporates refinements that address thermal, power delivery, and component reliability requirements identified through real-world enterprise deployments of the original ARS-111GL-NHR platform. Available through Omnixon Global with reach across UAE, GCC, EMEA, and APAC regions, this server is positioned as a purpose-built AI acceleration node for enterprises building or expanding GPU compute clusters in support of modern AI and high-performance computing initiatives.

Ideal for

• Large language model (LLM) inference serving for enterprise AI applications requiring low-latency, high-throughput GPU compute in a 1U form factor
• Generative AI workload acceleration, including text, image, and multimodal model deployment in production datacenter environments
• Scientific computing and simulation tasks such as molecular dynamics, computational fluid dynamics, and climate modeling leveraging Hopper GPU architecture
• High-density GPU cluster node deployment where rack space is constrained and maximum compute per rack unit is a primary design requirement
• Enterprise MLOps pipelines requiring dedicated GPU inference nodes integrated with standard PCIe-connected datacenter networking and NVMe storage fabrics
• Sovereign AI and regional cloud buildouts across EMEA and APAC where locally sourced, enterprise-grade GPU server hardware is required

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.