Confirm per delivered product ｜ AI GPUs · Professional Graphics · AI Servers & Systems ｜ 27 mainstream parts
NVIDIA AI product catalog covering GPU modules, data center GPUs, professional and consumer graphics cards, DGX servers, HGX platforms, liquid-cooled racks and desktop AI systems. Project inquiries only. Configuration, availability, pricing, stock, lead time, warranty and delivery regions are subject to project confirmation and quotation.
| Part No. | Series | Spec | Application | Quote |
|---|---|---|---|---|
| NVIDIA B300 SXM | AI GPU Compute Module | Blackwell Ultra architecture | LLM training & inference, agentic and reasoning models, real-time video generation · project-based quote | Quote → |
| NVIDIA B200 SXM | AI GPU Compute Module | Blackwell architecture, 180GB HBM3E | LLM training & inference, MoE models, data analytics, HPC · project-based quote | Quote → |
| NVIDIA Rubin GPU（SXM） | AI GPU Compute Module | Rubin architecture, 288GB HBM4 | Agentic AI, reasoning models, LLM training, high-concurrency inference, data analytics & HPC · project-based quote | Quote → |
| NVIDIA H200 SXM | AI GPU Compute Module | Hopper architecture, 141GB HBM3e | Generative AI, LLM training & inference, HPC, scientific computing · project-based quote | Quote → |
| NVIDIA H200 NVL | Data Center PCIe GPU | Hopper architecture, 141GB HBM3e | Enterprise LLM inference, generative AI, RAG, vision AI, scientific computing · project-based quote | Quote → |
| NVIDIA H100 SXM | AI GPU Compute Module | Hopper architecture, 80GB HBM3 | LLM training & inference, deep learning, HPC, data analytics, expanding existing AI clusters · project-based quote | Quote → |
| NVIDIA H100 PCIe 80GB | Data Center PCIe GPU | Hopper architecture, 80GB HBM2e | Enterprise AI training & inference, data analytics, HPC, expanding existing GPU servers · project-based quote | Quote → |
| NVIDIA RTX PRO 6000 Blackwell Server Edition | Data Center PCIe GPU | Blackwell architecture, 96GB GDDR7 with ECC | Enterprise LLM inference, agentic AI, industrial & physical AI, scientific computing, data analytics, 3D rendering, video processing · project-based quote | Quote → |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition | Professional Workstation GPU | Blackwell architecture, 96GB GDDR7 with ECC | Local LLM fine-tuning & inference, AI agent development, data science, engineering simulation, 3D rendering, video production · project-based quote | Quote → |
| NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition | Professional Workstation GPU | Blackwell architecture, 96GB GDDR7 with ECC | Multi-GPU workstations, local LLM fine-tuning & inference, AI agent development, data analytics, 3D rendering, complex simulation · project-based quote | Quote → |
| NVIDIA L40S | Data Center PCIe GPU | Ada Lovelace architecture, 48GB GDDR6 with ECC | Generative AI, LLM inference & training, 3D graphics, rendering, video processing, virtual workstations · project-based quote | Quote → |
| NVIDIA L4 Tensor Core GPU | Data Center PCIe GPU | Ada Lovelace architecture, 24GB GDDR6 with ECC | AI inference, video analytics, generative AI inference, graphics & virtual workstations, edge AI · project-based quote | Quote → |
| NVIDIA RTX 6000 Ada Generation | Professional Workstation GPU | Ada Lovelace architecture, 48GB GDDR6 with ECC | Professional 3D design, rendering, engineering simulation, AI development, data science, virtual workstations · project-based quote | Quote → |
| NVIDIA GeForce RTX 5090 | Consumer Graphics Card | Blackwell architecture, 32GB GDDR7 | Local AI model development & inference, generative image & video, 3D rendering, video editing, gaming · project-based quote | Quote → |
| NVIDIA GeForce RTX 4090 | Consumer Graphics Card | Ada Lovelace architecture, 24GB GDDR6X | Local AI development & inference, generative imaging, 3D rendering, video editing, gaming · project-based quote | Quote → |
| NVIDIA DGX B300 | AI Server System | Blackwell Ultra architecture, 8-GPU configuration | LLM training, inference, AI reasoning, enterprise AI clusters · project-based quote | Quote → |
| NVIDIA DGX B200 | AI Server System | Blackwell architecture, 8-GPU configuration, 1,440GB total GPU memory | LLM training, inference, data analytics, enterprise AI clusters · project-based quote | Quote → |
| NVIDIA DGX H200 | AI Server System | Hopper architecture, 8-GPU configuration, 1,128GB total GPU memory | LLM training, inference, data analytics, upgrading existing AI clusters · project-based quote | Quote → |
| NVIDIA DGX H100 | AI Server System | Hopper architecture, 8-GPU configuration, 640GB total GPU memory | AI training, inference, HPC, expanding existing clusters · project-based quote | Quote → |
| NVIDIA HGX B300 | AI GPU Platform | Blackwell Ultra architecture, 8-GPU configuration | LLM training, reasoning inference, OEM AI servers · project-based quote | Quote → |
| NVIDIA HGX B200 | AI GPU Platform | Blackwell architecture, 8-GPU configuration, 1,440GB total GPU memory | LLM training, inference, HPC, OEM AI servers · project-based quote | Quote → |
| NVIDIA HGX Rubin NVL8 | AI GPU Platform | Rubin architecture, 8-GPU configuration | LLM training, agentic AI inference, data analytics, HPC · project-based quote | Quote → |
| NVIDIA GB300 NVL72 | Liquid-cooled AI Rack Platform | Grace Blackwell Ultra architecture, 72-GPU configuration, ~20TB total GPU memory | Large AI factories, LLM training, reasoning inference, video generation · project-based quote | Quote → |
| NVIDIA GB200 NVL72 | Liquid-cooled AI Rack Platform | Grace Blackwell architecture, 72-GPU configuration | Large AI factories, LLM training, trillion-parameter model inference, HPC · project-based quote | Quote → |
| NVIDIA Vera Rubin NVL72 | Liquid-cooled AI Rack Platform | Vera Rubin architecture, 72-GPU configuration, ~20.7TB total GPU memory | Large AI factories, MoE model training, agentic AI inference, long context · project-based quote | Quote → |
| NVIDIA DGX Spark（128 GB） | Desktop AI Computer | GB10 Grace Blackwell architecture, 128GB coherent unified memory | Local LLM prototyping, fine-tuning, inference, AI agent development · project-based quote | Quote → |
| NVIDIA DGX Station（GB300） | Deskside AI Workstation Platform | GB300 Grace Blackwell Ultra Desktop Superchip, 748GB coherent unified memory | Local LLM development, fine-tuning, inference, data science, AI agents, physical AI · project-based quote | Quote → |
Send your BOM for NVIDIA pricing and alternatives — reply within 24 hours.