An enterprise-grade AI server engineered for training, fine-tuning, and high-throughput inference. Pack up to 8 NVIDIA Blackwell GPUs and scale confidently with expansive PCIe Gen5 I/O, redundant power, and datacenter-ready manageability.
Compute
Dual AMD EPYC™ 9005/9004/97x4
Accelerators
Up to 8× RTX 5090 / RTX PRO 6000 Blackwell
Memory
24× DDR5 ECC RDIMM
Power
5× 2000W
Networking
400 Gbps IB
Memory
24× DDR5 ECC RDIMM
2× AMD EPYC™ 9005/9004/97x4 processors deliver exceptional core density and memory bandwidth.
Eight PCIe Gen5 x16 slots support up to 8× NVIDIA RTX 5090 or RTX PRO 6000 Blackwell GPUs.
24× DDR5 ECC RDIMM slots keep large models and batches in-memory for faster throughput.
2× internal M.2 (PCIe 3.0x4) and 4× U.2 bays (PCIe 5.0x4) for blazing-fast datasets and checkpoints.
1× 400 Gbps InfiniBand, plus 2× PCIe 5.0×8 networking slots (mergeable to ×16) for scale-out clusters.
5× 2000 W PSUs for reliability under heavy mixed training/inference loads.
Everything you need for state-of-the-art generative AI workloads.
CPU: 2× AMD EPYC™ 9005, 9004 or 97x4 processors
Memory: 24× DDR5 ECC RDIMM slots
Expansion: 8× PCIe 5.0 ×16 GPU slots
Accelerators: Up to 8× NVIDIA RTX 5090 or RTX PRO 6000 Blackwell GPUs
2× Internal M.2 NVMe (PCIe 3.0 ×4)
4× U.2 NVMe bays (PCIe 5.0 ×4)
2× PCIe 5.0 ×8 slots for networking (mergeable to 1× PCIe 5.0 ×16)
1× 400 Gbps InfiniBand network card
Front I/O: IPMI port and 2× 1 GbE RJ45
5× 2000 W industrial-grade PSUs
Operating temperature: 10 °C – 35 °C
Non-operating temperature: −40 °C – 70 °C
Non-operating humidity: 20% – 90% (non-condensing)
1× DB15 VGA port
4× Type-A (USB 3.2 Gen1)
Dedicated IPMI for out-of-band management
Datacenter-ready chassis with optimized airflow and serviceability
Tool-assisted access for GPUs, storage, and power supplies
Your appliance arrives with DenseMAX Studio pre-installed for rapid time-to-value: project templates, model serving, fine-tuning pipelines, evaluation, guardrails, observability, and a collaborative data/model hub — all optimized for NVIDIA GPUs.
Launch from pre-built app templates or deploy custom services in minutes.
Production-grade serving: KV-cache routing, GPU sharding, replicas, and RBAC guardrails.
Training & fine-tuning workflows (full/LoRA), alignment (DPO/PPO/GRPO), quantization & distillation.
Data & Model Hub with Git-like ops, diffs, and HF/ModelScope import/export.
Chat, RAG, Agents, Evaluations
Metrics, traces, requests, guardrail events
Fine-tuning, alignment, distillation
Git-like ops for models & datasets
Request a guided demo or talk to our team about configurations, pricing, and delivery.
Enterprise AI that delivers: secure, sovereign-by-design, predictable, at scale. From POC to production with DenseMAX appliances, Digital Automation, and Surogate Agents.
