Enterprise AI Compute Kit NVIDIA
Tesla Referral V100 16GB SXM2 + PCIe Adapter + 11cm Cooling SXM2 to PCIe Conversion Kit for AI Deep Learning & HPC Why Buy This ✓ 16GB HBM2 high-bandwidth memory — for large AI models & high-performance computing ✓ SXM2 to PCIe adapter — use SXM2 GPU in standard PCIe server / workstation ✓ 11cm active cooling — solves SXM2 overheating in standard chassis ✓ Cost-effective — enterprise-grade AI compute at budget price What You'll Get
This is a complete SXM2-to-PCIe conversion kit built for AI and HPC workloads. It includes an original enterprise-pulled Tesla Referral V100 16GB SXM2 GPU, a brand-new SXM2 to PCIe adapter board and an 11cm active cooling solution, allowing you to install a high-performance SXM2 compute card into a standard PCIe x16 slot.
Powered by Tesla Referral V100 with Tensor cores and 16GB HBM2 memory, it delivers excellent performance for deep learning model training, AI inference, scientific computing, data processing and parallel computing tasks. Fully compatible with mainstream CUDA AI frameworks.
The adapter board and cooling kit are brand new. The GPU is pulled from decommissioned enterprise servers and fully tested. Assembly is required — you will need to mount the GPU onto the adapter board yourself.
Key Features
- NVIDIA Tesla Referral V100 16GB SXM2 compute GPU with Tensor Cores
- Brand new SXM2 to PCIe x16 adapter board for standard chassis
- 11cm active cooling fan included for reliable temperature control
- Supports CUDA, Tensor Core, cuDNN & mainstream AI frameworks
- Ideal for AI training, inference, HPC, rendering & data science
- Compatible with standard PCIe x16 slots in servers & workstations
- Supports 24/7 continuous operation under compute workloads
Specifications Item Detail GPU Model NVIDIA
Tesla Referral V100 16GB SXM2 Memory 16 GB HBM2 Host Interface PCIe x16 (via adapter board) Cooling 11cm active cooling fan kit GPU Condition Used, enterprise pull-out, fully tested Adapter & Cooling Brand new Display Output None — compute accelerator only Supported Software CUDA, cuDNN, PyTorch, TensorFlow, HPC frameworks Condition Mixed condition kit. The
Tesla Referral V100 SXM2 GPU is a used enterprise pull-unit; the SXM2-to-PCIe adapter board and cooling fan are brand new. All parts are fully tested before shipment.
- GPU passed full power-on, nvidia-smi verification & load testing
- Adapter board and cooling fan are brand new, unused
- GPU has normal server disassembly traces, no physical damage or burnt components
- PCIe gold pins on adapter are intact, no bending
- Assembly required: you need to mount the GPU onto the adapter board
- Packed in anti-static protective materials; no retail box
⚠ Important — Read Before Buying:
1. This is a compute-only accelerator with NO display outputs. It cannot be used as a gaming or display graphics card.
2. User assembly is required — you must install the SXM2 GPU onto the adapter board yourself.
3. Verify your chassis has enough space for the 11cm cooling assembly and your PSU provides sufficient power.
4. We are not responsible for incompatibility caused by insufficient power, space limits, or misunderstanding of product function. Package Includes
1 × NVIDIA Tesla Referral V100 16GB SXM2 GPU (used)
1 × Brand new SXM2 to PCIe adapter board
1 × 11cm active cooling fan kit
Anti-static protective packaging
Warranty & Returns Warranty 30-day functional warranty from delivery date Returns 30-day returns accepted (buyer pays return shipping if not DOA) Testing Every kit fully tested before dispatch for stable performance Shipping
- Dispatched within 1–3 business days after payment
- Tracking number provided for every order
- Standard & Expedited options at checkout
- International shipping via Global Shipping Program
Why Buy From Us
- Real stock — what you see is what ships
- Full testing — every GPU and adapter tested before shipment
- Secure packaging — anti-static & shockproof protection
- Fast response — message us anytime before or after purchase
FAQ
Q: Does it come pre-assembled?
A: No. You need to mount the SXM2 GPU onto the adapter board yourself.
Q: Can I use this for gaming?
A: No. It has no display outputs and is a compute-only accelerator.
Q: Is the adapter board new or used?
A: The adapter board and cooling fan are brand new; the GPU is a tested used enterprise pull.
Q: What software does it support?
A: It supports CUDA, cuDNN, PyTorch, TensorFlow and all mainstream AI & HPC frameworks.
Questions about compatibility or assembly? Message us before ordering — we reply fast.
