{"product_id":"nvidia-h200-nvl-141gb-hbm3e-pcie-gpu","title":"NVIDIA H200 NVL 141GB HBM3e PCIe GPU","description":"\u003caside class=\"phd-note\" id=\"purchase-quantity\"\u003e\u003ch3\u003eOrdering more than one card?\u003c\/h3\u003e\n\u003cp\u003eOne card of this model per customer. Need more than one? \u003ca style=\"color: inherit;\" href=\"https:\/\/perceptasolutions.com\/hardware-compute?topic=hardware\u0026amp;product_title=NVIDIA+H200+NVL+141GB+HBM3e+PCIe+GPU\u0026amp;product_handle=nvidia-h200-nvl-141gb-hbm3e-pcie-gpu#contact\"\u003eContact us to discuss a multi-card order ↗\u003c\/a\u003e\u003c\/p\u003e\u003c\/aside\u003e\n\u003csection class=\"phd-intro\"\u003e\n  \u003cdiv\u003e\n\u003cp class=\"phd-eyebrow\"\u003eData centre GPU \/ NVIDIA Hopper\u003c\/p\u003e\n\u003ch2\u003e141GB of HBM3e for AI and scientific computing.\u003c\/h2\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cp\u003eThe NVIDIA H200 NVL pairs 141GB of HBM3e with 4.8TB\/s memory bandwidth for LLM inference, model training and memory-intensive research. Its Hopper architecture combines Tensor Core acceleration for AI with FP64 compute for scientific applications, in a dual-slot PCIe server card.\u003c\/p\u003e\n\u003cp\u003eThis new H200 NVL PCIe accelerator (900-21010-0040-000) requires a compatible server with suitable power and forced airflow.\u003c\/p\u003e\n\u003c\/div\u003e\n\u003c\/section\u003e\n\u003cnav class=\"phd-jump-links\" aria-label=\"Product information\"\u003e\n  \u003ca href=\"#hardware-components\"\u003eWorkloads\u003c\/a\u003e\n  \u003ca href=\"#hardware-specification\"\u003eTechnical specification\u003c\/a\u003e\n  \u003ca href=\"#hardware-included\"\u003eSystem requirements\u003c\/a\u003e\n  \u003ca href=\"#hardware-questions\"\u003eBefore you order\u003c\/a\u003e\n\u003c\/nav\u003e\n\u003cdiv class=\"phd-metrics\" aria-label=\"Key specifications\"\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003e141GB\u003c\/strong\u003e\u003cspan\u003eHBM3e memory per GPU\u003c\/span\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003e4.8TB\/s\u003c\/strong\u003e\u003cspan\u003eGPU memory bandwidth\u003c\/span\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003ePCIe 5.0\u003c\/strong\u003e\u003cspan\u003ex16 host interface\u003c\/span\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cstrong\u003eUp to 600W\u003c\/strong\u003e\u003cspan\u003eConfigurable GPU TDP\u003c\/span\u003e\n\u003c\/div\u003e\n\u003c\/div\u003e\n\u003csection class=\"phd-components\" id=\"hardware-components\"\u003e\n  \u003ch2\u003eAI inference, training and scientific computing\u003c\/h2\u003e\n  \u003cdiv class=\"phd-features\"\u003e\n    \u003carticle\u003e\u003ch3\u003eLLM inference and model training\u003c\/h3\u003e\n\u003cp\u003eHigh-capacity HBM3e provides space for model weights and runtime data, while Hopper Tensor Cores support mixed-precision inference and training. Inference memory requirements include context and concurrent requests, not just the model itself. Supported NVLink configurations allow applications designed for multiple GPUs to scale across cards.\u003c\/p\u003e\u003c\/article\u003e\n    \u003carticle\u003e\u003ch3\u003eSimulation and numerical research\u003c\/h3\u003e\n\u003cp\u003eFP64 and FP64 Tensor Core capabilities serve GPU-accelerated numerical applications that need double precision. The 4.8TB\/s memory bandwidth is particularly relevant to simulations and scientific analysis that repeatedly move large working datasets through GPU memory.\u003c\/p\u003e\u003c\/article\u003e\n    \u003carticle\u003e\u003ch3\u003eSeparate workloads on one GPU\u003c\/h3\u003e\n\u003cp\u003eMIG can partition the H200 NVL into up to seven isolated instances, letting smaller inference services or research jobs share the accelerator. Available profiles depend on the driver, software stack and host platform.\u003c\/p\u003e\u003c\/article\u003e\n  \u003c\/div\u003e\n\u003c\/section\u003e\n\u003caside class=\"phd-note\"\u003e\u003ch3\u003ePassive cooling means server-provided airflow.\u003c\/h3\u003e\n\u003cp\u003eThe passive heatsink needs sufficient ducted airflow from the server. Check the system manufacturer's H200 NVL support, power delivery and mechanical clearance before installation.\u003c\/p\u003e\u003c\/aside\u003e\n\u003csection class=\"phd-specs\" id=\"hardware-specification\"\u003e\n  \u003cdiv class=\"phd-section-heading\"\u003e\n\u003ch2\u003eTechnical specification\u003c\/h2\u003e\n\u003cp\u003eSpecifications per GPU unless stated\u003c\/p\u003e\n\u003c\/div\u003e\n  \u003ctable\u003e\n\u003ccaption\u003e01 \/ Architecture and memory\u003c\/caption\u003e\n\u003ctbody\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eGPU\u003c\/th\u003e\n\u003ctd\u003eNVIDIA H200 NVL\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eArchitecture\u003c\/th\u003e\n\u003ctd\u003eNVIDIA Hopper\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eMemory\u003c\/th\u003e\n\u003ctd\u003e141GB HBM3e\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eMemory bandwidth\u003c\/th\u003e\n\u003ctd\u003e4.8TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eMulti-Instance GPU\u003c\/th\u003e\n\u003ctd\u003eUp to 7 instances; supported profiles depend on the software and platform\u003c\/td\u003e\n\u003c\/tr\u003e\n  \u003ctr\u003e\n\u003cth scope=\"row\"\u003eDecode engines\u003c\/th\u003e\n\u003ctd\u003e7 NVDEC and 7 JPEG engines\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n  \u003ctable\u003e\n\u003ccaption\u003e02 \/ Peak compute performance\u003c\/caption\u003e\n\u003ctbody\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP64\u003c\/th\u003e\n\u003ctd\u003e30 TFLOPS\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP64 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e60 TFLOPS\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP32\u003c\/th\u003e\n\u003ctd\u003e60 TFLOPS\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eTF32 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e835 TFLOPS with sparsity\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eBF16 \/ FP16 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e1,671 TFLOPS with sparsity\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eFP8 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e3,341 TFLOPS with sparsity\u003c\/td\u003e\n\u003c\/tr\u003e\n  \u003ctr\u003e\n\u003cth scope=\"row\"\u003eINT8 Tensor Core\u003c\/th\u003e\n\u003ctd\u003e3,341 TOPS with sparsity\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/tbody\u003e\n\u003c\/table\u003e\n  \u003ctable\u003e\n\u003ccaption\u003e03 \/ Connectivity and multi-GPU support\u003c\/caption\u003e\n\u003ctbody\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eHost interface\u003c\/th\u003e\n\u003ctd\u003ePCIe 5.0 x16\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003ePCIe bandwidth\u003c\/th\u003e\n\u003ctd\u003e128GB\/s total bidirectional\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eGPU interconnect\u003c\/th\u003e\n\u003ctd\u003eNVIDIA NVLink, up to 900GB\/s per GPU total bidirectional\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eNVLink configurations\u003c\/th\u003e\n\u003ctd\u003eSupported 2- or 4-GPU bridge configurations; compatible server layout and bridge hardware required\u003c\/td\u003e\n\u003c\/tr\u003e\n  \u003c\/tbody\u003e\n\u003c\/table\u003e\n  \u003ctable\u003e\n\u003ccaption\u003e04 \/ Power, cooling and physical format\u003c\/caption\u003e\n\u003ctbody\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eMaximum TDP\u003c\/th\u003e\n\u003ctd\u003eUp to 600W, configurable within supported platform limits\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eCooling\u003c\/th\u003e\n\u003ctd\u003ePassive air-cooled heatsink; forced chassis airflow required\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eForm factor\u003c\/th\u003e\n\u003ctd\u003eFull-height, full-length, dual-slot PCIe card\u003c\/td\u003e\n\u003c\/tr\u003e\n    \u003ctr\u003e\n\u003cth scope=\"row\"\u003eNominal card dimensions\u003c\/th\u003e\n\u003ctd\u003e4.4 inches high x 10.5 inches long; dual-slot width\u003c\/td\u003e\n\u003c\/tr\u003e\n  \u003c\/tbody\u003e\n\u003c\/table\u003e\n  \u003cp class=\"phd-footnote\"\u003eCompute figures are peak values; results depend on the application and server configuration. Sparse performance requires supported sparse operations. Bidirectional bandwidth is the total across both directions, and TDP applies to the GPU alone. Allow additional installation space for cabling and airflow. NVIDIA lists the H200 specifications as preliminary and subject to change; confirm the board specifications before purchase.\u003c\/p\u003e\n\u003c\/section\u003e\n\u003csection class=\"phd-supply\" id=\"hardware-included\"\u003e\n  \u003cdiv\u003e\n\u003cp class=\"phd-eyebrow\"\u003eGPU supply\u003c\/p\u003e\n\u003ch2\u003eH200 NVL accelerator\u003c\/h2\u003e\n\u003cp\u003eThe listing is for the H200 NVL accelerator. A server and additional GPUs are not included.\u003c\/p\u003e\n\u003cp\u003eThe card is new and includes a 24-month Percepta return-to-base hardware warranty from delivery. A server, NVLink bridges, additional power cables and installation services are not included. NVIDIA AI Enterprise entitlement is subject to NVIDIA activation and system requirements; contact us about licensing for your deployment.\u003c\/p\u003e\n\u003c\/div\u003e\n  \u003cdiv\u003e\n\u003cp class=\"phd-eyebrow\"\u003ePlan your installation\u003c\/p\u003e\n\u003ch2\u003eWhat your server needs.\u003c\/h2\u003e\n\u003cul\u003e\n\u003cli\u003eExplicit H200 NVL support from the system manufacturer, with appropriate firmware and a supported PCIe slot or riser.\u003c\/li\u003e\n\u003cli\u003eChassis airflow and power delivery rated for the intended GPU configuration, including approved auxiliary power cabling.\u003c\/li\u003e\n\u003cli\u003eMechanical clearance for the card, cabling and any NVLink bridges.\u003c\/li\u003e\n\u003cli\u003eA supported operating system, NVIDIA driver and application stack; additional software licences where required.\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003c\/div\u003e\n\u003c\/section\u003e\n\u003csection class=\"phd-faq\" id=\"hardware-questions\" aria-label=\"Before you order\"\u003e\n  \u003cdetails\u003e\u003csummary\u003eWill H200 NVL work in my server?\u003c\/summary\u003e\u003cdiv\u003e\n\u003cp\u003eCheck explicit H200 NVL support with the server manufacturer, including firmware, PCIe slot or riser, auxiliary power cabling, cooling and mechanical clearance. The passive heatsink requires forced chassis airflow, and GPU TDP is configurable up to 600W within supported platform limits.\u003c\/p\u003e\n\u003cp\u003eH200 NVL is a PCIe card and is not interchangeable with an H200 SXM module. Send us the server make and model, riser and power details, and intended software environment to discuss the installation.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eHow do I check whether an AI model will fit?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eMemory use depends on the model, precision, context length, batch size, concurrency and serving software. Include working buffers and runtime overhead when assessing the 141GB capacity. Send us these workload details to discuss a suitable configuration.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eCan the GPU be divided between workloads?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eMIG supports up to seven GPU instances. The available profiles and setup requirements depend on the driver, software release and host platform. Check the supported profile table for your intended configuration.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eCan I connect several H200 NVL cards?\u003c\/summary\u003e\u003cdiv\u003e\n\u003cp\u003eSuitable systems support two- or four-GPU NVLink bridge configurations. Check card spacing, bridge hardware, power, cooling and software support with the system manufacturer.\u003c\/p\u003e\n\u003cp\u003eEach GPU has its own memory; applications must support distributing work across cards to use them together. Confirm any additional cards and bridges separately from this listing.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eIs NVIDIA AI Enterprise included?\u003c\/summary\u003e\u003cdiv\u003e\u003cp\u003eNVIDIA advertises a five-year AI Enterprise subscription for eligible H200 NVL products. Contact us to confirm entitlement, activation eligibility and system requirements for the card being supplied. The software subscription is separate from the hardware warranty.\u003c\/p\u003e\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eWhat comes with the accelerator?\u003c\/summary\u003e\u003cdiv\u003e\n\u003cp\u003eThe listing covers one new H200 NVL card, part number 900-21010-0040-000, rather than a complete server or multi-GPU package. It includes a 24-month Percepta return-to-base hardware warranty from delivery.\u003c\/p\u003e\n\u003cp\u003eDiscuss any NVLink bridges, auxiliary power cabling or installation service required for your server so these can be agreed as part of the supply.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\n  \n  \n  \n  \n  \u003cdetails\u003e\u003csummary\u003eWhen will my pre-order arrive, and can I cancel?\u003c\/summary\u003e\u003cdiv\u003e\n\u003cp\u003eThe supplier currently estimates availability on 1 January 2027, subject to NVIDIA supply. This is not a dispatch or delivery date. The agreed delivery deadline is 31 January 2027 unless you expressly agree an extension; otherwise we will cancel and refund in full.\u003c\/p\u003e\n\u003cp\u003eFull payment is taken at checkout. Both business and retail customers can cancel before dispatch for a full refund. Read the \u003ca href=\"\/pages\/pre-order-terms\"\u003epre-order terms\u003c\/a\u003e for delivery, cancellation and refund details.\u003c\/p\u003e\n\u003c\/div\u003e\u003c\/details\u003e\u003cdetails\u003e\u003csummary\u003eManufacturer documentation\u003c\/summary\u003e\u003cdiv\u003e\n\n\u003cul\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-us\/data-center\/h200\/\"\u003eNVIDIA H200 product specifications - H200 NVL column\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/dam-cdn.nvd.orangelogic.com\/AssetLink\/5o2qgy5d2835ve2pm11i62kv8mphqta8.pdf\"\u003eNVIDIA H200 Tensor Core GPU datasheet\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.pny.com\/File%20Library\/Company\/Support\/linecards\/data-center-gpus\/nvidia-data-center-gpu-linecard.pdf\"\u003ePNY data centre GPU linecard - physical format\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/docs.nvidia.com\/datacenter\/tesla\/mig-user-guide\/supported-mig-profiles.html\"\u003eNVIDIA supported MIG profiles\u003c\/a\u003e\u003c\/li\u003e\n\u003cli\u003e\u003ca href=\"https:\/\/www.nvidia.com\/en-us\/data-center\/activate-license\/\"\u003eNVIDIA AI Enterprise entitlement and activation guidance\u003c\/a\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003c\/div\u003e\u003c\/details\u003e\n\u003c\/section\u003e\n","brand":"NVIDIA","offers":[{"title":"Default Title","offer_id":60871535264078,"sku":"900-21010-0040-000","price":33995.0,"currency_code":"GBP","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/1078\/0213\/2814\/files\/h200-nvl-3qtr-right-reverse.png?v=1789840138","url":"https:\/\/shop.perceptasolutions.com\/products\/nvidia-h200-nvl-141gb-hbm3e-pcie-gpu","provider":"Percepta Solutions","version":"1.0","type":"link"}