Nvidia CPU Servers and Accelerators –

Enterprise System Pulls (tested working)

NVIDIA GPU Accelerators

We stock high-performance NVIDIA GPUs for AI training, inference, and high-density computing. All units are data-center pulled or new-old-stock, fully tested, and ready for deployment.

Current High-Demand Models

RTX PRO 6000 Blackwell Server Edition (PNY) 96 GB GDDR7 ECC | up to 1.6 TB/s | 24,064 CUDA Cores | PCIe 5.0 x16 | up to 600W

Sample 8× Configs:

  • Supermicro 4U/6U MGX → 8× RTX PRO 6000 Blackwell + dual AMD EPYC 9005 / Intel Xeon 6 + 1–2 TB DDR5 + high-speed networking

  • ASUS ESC8000 / MGX → 8× RTX PRO 6000 Blackwell + dual EPYC or Xeon + 1.5 TB+ DDR5

  • Dell / Lenovo 4U–6U → 8× RTX PRO 6000 Blackwell configurations available

NVIDIA H100 80 GB HBM3 (SXM) / 94 GB (NVL)

Sample 8× Configs:

  • Supermicro SYS-821GE-TNHR → 8× H100 SXM + dual AMD EPYC or Intel Xeon + up to 2 TB DDR5 + 8× NVMe + InfiniBand options

  • Dell PowerEdge XE9680 → 8× H100 SXM + dual Xeon / EPYC + 1.5–2 TB DDR5 + enterprise networking

  • ASUS / Lenovo 8-GPU platforms → Fully loaded H100 SXM or NVL configurations

NVIDIA H200 141 GB HBM3e | 4.8 TB/s

Sample 8× Configs:

  • Supermicro SYS-821GE / ARS series → 8× H200 SXM + dual high-core EPYC or Xeon 6 + 2 TB DDR5 + liquid or air cooling

  • Dell PowerEdge XE9680 → 8× H200 SXM + dual processors + 1.5–2 TB DDR5 + full NVLink fabric

  • ASUS MGX / Lenovo → 8× H200 configurations with high memory density

NVIDIA B300 (Blackwell Ultra) 288 GB HBM3e | High FP4 performance

Sample 8× Configs:

  • Supermicro HGX B300 / ARS platforms → 8× B300 + dual AMD EPYC 9005/9006 or Intel Xeon 6 + 2 TB+ DDR5 + direct liquid cooling options

  • Dell PowerEdge XE9712 → 8× B300 + high-core CPUs + large DDR5 capacity + advanced cooling

  • ASUS / Lenovo Blackwell Ultra systems → Fully loaded 8× B300 nodes available

Typical Fully Loaded Specs (8× GPU nodes)

  • GPUs: 8× H100 / H200 / B300 or RTX PRO 6000 Blackwell

  • CPUs: Dual AMD EPYC 9005/9006 or Intel Xeon 6 (64–192 cores)

  • Memory: 1.5 TB – 2 TB+ DDR5 ECC RDIMM

  • Storage: Multiple NVMe U.2 / E1.S drives

  • Networking: 400G / 800G InfiniBand or Ethernet options

  • Cooling: Air or Direct Liquid Cooling (DLC)

All systems are tested and ready to ship. Contact us for current stock, bulk pricing, or fully configured AI servers.

Nvidia enterprise pulls are better.