{"product_id":"nvda-business-model-canvas","title":"NVIDIA Corporation (NVDA): Business Model Canvas [June-2026 Updated]","description":"\u003cp\u003eThis ready-made Business Model Canvas gives you a practical, research-based view of how NVIDIA Corporation turns advanced chip design, CUDA software, and AI infrastructure into revenue across data center, edge, enterprise AI, and AI PC markets. You'll see how partnerships with TSMC, AWS, Google Cloud, Microsoft Azure, Oracle, Dell, HPE, Lenovo, Supermicro, and Groq support 3nm, CoWoS, and HBM4 supply, while value drivers like up to \u003cstrong\u003e10x\u003c\/strong\u003e lower MoE cost per token, up to \u003cstrong\u003e7x\u003c\/strong\u003e faster inference, and local \u003cstrong\u003e1-trillion-parameter\u003c\/strong\u003e AI on DGX Station shape demand from hyperscalers, frontier AI labs, large enterprises, and OEM buyers. It also breaks down the main cost pressures, including R\u0026amp;D, advanced packaging, inventory write-downs, and export-control risk, so you can quickly understand the operating model, strategic resources, and revenue logic behind a company with a \u003cstrong\u003e$5.42T\u003c\/strong\u003e market value.\u003c\/p\u003e\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Key Partnerships\u003c\/h2\u003e\n\u003cp\u003eNVIDIA reported \u003cstrong\u003e$26.044B\u003c\/strong\u003e revenue in Q1 FY2025 and \u003cstrong\u003e$22.6B\u003c\/strong\u003e Data Center revenue. The partnership base is centered on \u003cstrong\u003e3nm\u003c\/strong\u003e chipmaking, \u003cstrong\u003eCoWoS\u003c\/strong\u003e packaging, and \u003cstrong\u003e8\u003c\/strong\u003e-GPU system blocks.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eTSMC for 3nm and CoWoS capacity\u003c\/strong\u003e\u003c\/p\u003e\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003ePartner\u003c\/td\u003e\n\u003ctd\u003eNode\u003c\/td\u003e\n\u003ctd\u003ePackaging\u003c\/td\u003e\n\u003ctd\u003ePublic NVIDIA-specific capacity figure\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTSMC\u003c\/td\u003e\n\u003ctd\u003e3nm\u003c\/td\u003e\n\u003ctd\u003eCoWoS\u003c\/td\u003e\n\u003ctd\u003e0\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eVera Rubin\u003c\/td\u003e\n\u003ctd\u003e2026\u003c\/td\u003e\n\u003ctd\u003eCoWoS\u003c\/td\u003e\n\u003ctd\u003eNot disclosed\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003e3nm\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003eCoWoS\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e0\u003c\/strong\u003e public NVIDIA-specific CoWoS capacity figure\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eAWS, Google Cloud, Microsoft Azure, Oracle\u003c\/strong\u003e\u003c\/p\u003e\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003ePartner\u003c\/td\u003e\n\u003ctd\u003eInstance\u003c\/td\u003e\n\u003ctd\u003eGPUs\u003c\/td\u003e\n\u003ctd\u003eTotal H100 GPU memory\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAWS\u003c\/td\u003e\n\u003ctd\u003ep5.48xlarge\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGoogle Cloud\u003c\/td\u003e\n\u003ctd\u003eA3\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eMicrosoft Azure\u003c\/td\u003e\n\u003ctd\u003eND H100 v5\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOracle\u003c\/td\u003e\n\u003ctd\u003eBM.GPU.H100.8\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\u003cul\u003e\n\u003cli\u003eAWS: \u003cstrong\u003e8\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003eGoogle Cloud: \u003cstrong\u003e8\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003eMicrosoft Azure: \u003cstrong\u003e8\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003eOracle: \u003cstrong\u003e8\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eDell, HPE, Lenovo, Supermicro\u003c\/strong\u003e\u003c\/p\u003e\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003ePartner\u003c\/td\u003e\n\u003ctd\u003eSystem\u003c\/td\u003e\n\u003ctd\u003eGPUs\u003c\/td\u003e\n\u003ctd\u003eTotal H100 GPU memory\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDell\u003c\/td\u003e\n\u003ctd\u003ePowerEdge XE9680\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eHPE\u003c\/td\u003e\n\u003ctd\u003eCray XD670\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLenovo\u003c\/td\u003e\n\u003ctd\u003eThinkSystem SR675 V3\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSupermicro\u003c\/td\u003e\n\u003ctd\u003e8-GPU systems\u003c\/td\u003e\n\u003ctd\u003e8\u003c\/td\u003e\n\u003ctd\u003e640 GB\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e8\u003c\/strong\u003e-GPU servers across all 4 partners\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e640 GB\u003c\/strong\u003e per 8 x 80 GB H100 node\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eGroq technology licensing partnership\u003c\/strong\u003e\u003c\/p\u003e\n\u003cp\u003e\u003cstrong\u003e0\u003c\/strong\u003e public disclosed NVIDIA licensing fee, \u003cstrong\u003e0\u003c\/strong\u003e public disclosed royalty rate, \u003cstrong\u003e0\u003c\/strong\u003e public disclosed unit volume.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eTaiwan ecosystem partners for Vera Rubin\u003c\/strong\u003e\u003c\/p\u003e\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003ePartner\u003c\/td\u003e\n\u003ctd\u003eRole\u003c\/td\u003e\n\u003ctd\u003eNumeric detail\u003c\/td\u003e\n\u003ctd\u003ePlatform\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eTSMC\u003c\/td\u003e\n\u003ctd\u003eFoundry and packaging\u003c\/td\u003e\n\u003ctd\u003e3nm\u003c\/td\u003e\n\u003ctd\u003eVera Rubin\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFoxconn\u003c\/td\u003e\n\u003ctd\u003eServer assembly\u003c\/td\u003e\n\u003ctd\u003e8-GPU\u003c\/td\u003e\n\u003ctd\u003eVera Rubin\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eQuanta\u003c\/td\u003e\n\u003ctd\u003eServer assembly\u003c\/td\u003e\n\u003ctd\u003e8-GPU\u003c\/td\u003e\n\u003ctd\u003eVera Rubin\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWistron\u003c\/td\u003e\n\u003ctd\u003eServer assembly\u003c\/td\u003e\n\u003ctd\u003e8-GPU\u003c\/td\u003e\n\u003ctd\u003eVera Rubin\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eWiwynn\u003c\/td\u003e\n\u003ctd\u003eServer assembly\u003c\/td\u003e\n\u003ctd\u003e8-GPU\u003c\/td\u003e\n\u003ctd\u003eVera Rubin\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e3nm\u003c\/strong\u003e at TSMC\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eCoWoS\u003c\/strong\u003e at TSMC\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e8\u003c\/strong\u003e-GPU server assembly in Taiwan\u003c\/li\u003e\n\u003c\/ul\u003e\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Key Activities\u003c\/h2\u003e\n\u003cp\u003e\u003cstrong\u003eNVIDIA Corporation's key activities are built around 208 billion-transistor Blackwell GPUs, 80 billion-transistor Hopper GPUs, 72-core Grace CPUs, rack-scale systems with 72 GPUs and 36 CPUs, and a software stack that runs from CUDA to TensorRT-LLM and Triton Inference Server.\u003c\/strong\u003e\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003cth\u003eKey activity\u003c\/th\u003e\n\u003cth\u003eReal-life products or systems\u003c\/th\u003e\n\u003cth\u003eNumeric anchors\u003c\/th\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDesign AI chips and CPUs\u003c\/td\u003e\n\u003ctd\u003eBlackwell B200, H200, H100, Grace CPU, GB200 NVL72\u003c\/td\u003e\n\u003ctd\u003e208 billion transistors; 192 GB HBM3e; 8 TB\/s; 141 GB HBM3e; 4.8 TB\/s; 80 billion transistors; 80 GB HBM3; 3.35 TB\/s; 72 GPUs; 36 CPUs\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDevelop inference software and AI stacks\u003c\/td\u003e\n\u003ctd\u003eCUDA, cuDNN, cuBLAS, NCCL, TensorRT, TensorRT-LLM, Triton Inference Server, NeMo, NIM\u003c\/td\u003e\n\u003ctd\u003e9 named software layers across training and inference\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eIntegrate AI factory hardware and applications\u003c\/td\u003e\n\u003ctd\u003eDGX H100, DGX B200, BlueField-3, Spectrum-X\u003c\/td\u003e\n\u003ctd\u003e8 GPUs; 640 GB HBM3; 1,536 GB HBM3e; 400 Gb\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSecure advanced packaging and HBM supply\u003c\/td\u003e\n\u003ctd\u003eTSMC advanced packaging; HBM3; HBM3e\u003c\/td\u003e\n\u003ctd\u003e80 GB; 141 GB; 192 GB; 3.35 TB\/s; 4.8 TB\/s; 8 TB\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eSupport cloud and OEM deployments\u003c\/td\u003e\n\u003ctd\u003eGB200 NVL72, DGX systems, OEM server partners\u003c\/td\u003e\n\u003ctd\u003e72 GPUs; 36 CPUs; 8-GPU nodes; 400 Gb\/s\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003cth\u003ePeriod ended\u003c\/th\u003e\n\u003cth\u003eTotal revenue\u003c\/th\u003e\n\u003cth\u003eData center revenue\u003c\/th\u003e\n\u003cth\u003eOther revenue\u003c\/th\u003e\n\u003cth\u003eData center share\u003c\/th\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFY2025\u003c\/td\u003e\n\u003ctd\u003e$130.5B\u003c\/td\u003e\n\u003ctd\u003e$115.2B\u003c\/td\u003e\n\u003ctd\u003e$15.3B\u003c\/td\u003e\n\u003ctd\u003e88.3%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eQ4 FY2025\u003c\/td\u003e\n\u003ctd\u003e$39.3B\u003c\/td\u003e\n\u003ctd\u003e$35.6B\u003c\/td\u003e\n\u003ctd\u003e$3.7B\u003c\/td\u003e\n\u003ctd\u003e90.6%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003eDesign AI chips and CPUs\u003c\/strong\u003e\u003c\/p\u003e\n\u003cp\u003eBlackwell B200 carries \u003cstrong\u003e208 billion\u003c\/strong\u003e transistors, \u003cstrong\u003e192 GB\u003c\/strong\u003e of HBM3e, and \u003cstrong\u003e8 TB\/s\u003c\/strong\u003e of memory bandwidth. H200 carries \u003cstrong\u003e141 GB\u003c\/strong\u003e of HBM3e and \u003cstrong\u003e4.8 TB\/s\u003c\/strong\u003e. H100 carries \u003cstrong\u003e80 billion\u003c\/strong\u003e transistors, \u003cstrong\u003e80 GB\u003c\/strong\u003e of HBM3, and \u003cstrong\u003e3.35 TB\/s\u003c\/strong\u003e. GB200 combines \u003cstrong\u003e1\u003c\/strong\u003e Grace CPU with \u003cstrong\u003e2\u003c\/strong\u003e Blackwell GPUs, and GB200 NVL72 scales that to \u003cstrong\u003e36\u003c\/strong\u003e Grace CPUs and \u003cstrong\u003e72\u003c\/strong\u003e Blackwell GPUs. That scale is the core of NVIDIA Corporation's hardware roadmap.\u003c\/p\u003e\n\n\u003cul\u003e\n\u003cli\u003eBlackwell B200: \u003cstrong\u003e208 billion\u003c\/strong\u003e transistors, \u003cstrong\u003e192 GB\u003c\/strong\u003e HBM3e, \u003cstrong\u003e8 TB\/s\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003eH200: \u003cstrong\u003e141 GB\u003c\/strong\u003e HBM3e, \u003cstrong\u003e4.8 TB\/s\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003eH100: \u003cstrong\u003e80 billion\u003c\/strong\u003e transistors, \u003cstrong\u003e80 GB\u003c\/strong\u003e HBM3, \u003cstrong\u003e3.35 TB\/s\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003eGB200 NVL72: \u003cstrong\u003e72\u003c\/strong\u003e GPUs, \u003cstrong\u003e36\u003c\/strong\u003e CPUs\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eDevelop inference software and AI stacks\u003c\/strong\u003e\u003c\/p\u003e\n\u003cp\u003eThe inference stack spans \u003cstrong\u003e9\u003c\/strong\u003e named layers: CUDA, cuDNN, cuBLAS, NCCL, TensorRT, TensorRT-LLM, Triton Inference Server, NeMo, and NIM. This matters because the chip sale is only the first step; the software stack is what lets NVIDIA Corporation's hardware run models after training, serve prompts, and stay inside customer workflows.\u003c\/p\u003e\n\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e9\u003c\/strong\u003e software layers across training and inference\u003c\/li\u003e\n\u003cli\u003eCUDA as the base programming layer\u003c\/li\u003e\n\u003cli\u003eTensorRT-LLM for large language model inference\u003c\/li\u003e\n\u003cli\u003eTriton Inference Server for model serving\u003c\/li\u003e\n\u003cli\u003eNeMo and NIM for deployment and application delivery\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eIntegrate AI factory hardware and applications\u003c\/strong\u003e\u003c\/p\u003e\n\u003cp\u003eNVIDIA Corporation sells systems as well as chips. DGX H100 uses \u003cstrong\u003e8\u003c\/strong\u003e H100 GPUs and \u003cstrong\u003e640 GB\u003c\/strong\u003e of total HBM3 memory. DGX B200 uses \u003cstrong\u003e8\u003c\/strong\u003e B200 GPUs and \u003cstrong\u003e1,536 GB\u003c\/strong\u003e of total HBM3e memory. BlueField-3 adds \u003cstrong\u003e400 Gb\/s\u003c\/strong\u003e networking capacity, which matters because AI factories need compute, storage, networking, and security in the same rack. The system work turns individual accelerators into deployable infrastructure.\u003c\/p\u003e\n\n\u003cul\u003e\n\u003cli\u003eDGX H100: \u003cstrong\u003e8\u003c\/strong\u003e H100 GPUs, \u003cstrong\u003e640 GB\u003c\/strong\u003e total HBM3\u003c\/li\u003e\n\u003cli\u003eDGX B200: \u003cstrong\u003e8\u003c\/strong\u003e B200 GPUs, \u003cstrong\u003e1,536 GB\u003c\/strong\u003e total HBM3e\u003c\/li\u003e\n\u003cli\u003eBlueField-3: \u003cstrong\u003e400 Gb\/s\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003eGB200 NVL72: \u003cstrong\u003e72\u003c\/strong\u003e GPUs, \u003cstrong\u003e36\u003c\/strong\u003e CPUs\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eSecure advanced packaging and HBM supply\u003c\/strong\u003e\u003c\/p\u003e\n\u003cp\u003eHBM, high-bandwidth memory, is a major constraint in NVIDIA Corporation's supply chain. The move from H100's \u003cstrong\u003e80 GB\u003c\/strong\u003e to H200's \u003cstrong\u003e141 GB\u003c\/strong\u003e and B200's \u003cstrong\u003e192 GB\u003c\/strong\u003e shows how much memory content has increased. That shift raises dependence on HBM3e stacks, advanced packaging, and substrate capacity from partners such as TSMC, SK hynix, Micron, and Samsung. Without those inputs, the hardware roadmap cannot scale.\u003c\/p\u003e\n\n\u003cul\u003e\n\u003cli\u003eH100: \u003cstrong\u003e80 GB\u003c\/strong\u003e HBM3\u003c\/li\u003e\n\u003cli\u003eH200: \u003cstrong\u003e141 GB\u003c\/strong\u003e HBM3e\u003c\/li\u003e\n\u003cli\u003eB200: \u003cstrong\u003e192 GB\u003c\/strong\u003e HBM3e\u003c\/li\u003e\n\u003cli\u003eMemory bandwidth moves from \u003cstrong\u003e3.35 TB\/s\u003c\/strong\u003e to \u003cstrong\u003e4.8 TB\/s\u003c\/strong\u003e to \u003cstrong\u003e8 TB\/s\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eSupport cloud and OEM deployments\u003c\/strong\u003e\u003c\/p\u003e\n\u003cp\u003eCloud and OEM support depends on taking the same architecture to \u003cstrong\u003e8-GPU\u003c\/strong\u003e nodes and \u003cstrong\u003e72-GPU\u003c\/strong\u003e racks. GB200 NVL72 pairs \u003cstrong\u003e36\u003c\/strong\u003e Grace CPUs with \u003cstrong\u003e72\u003c\/strong\u003e Blackwell GPUs, which is why deployment work covers rack design, power, cooling, networking, drivers, firmware, and qualification. BlueField-3 at \u003cstrong\u003e400 Gb\/s\u003c\/strong\u003e sits inside that deployment model because the AI factory has to move data as fast as it computes it.\u003c\/p\u003e\n\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e8-GPU\u003c\/strong\u003e deployment nodes\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e72-GPU\u003c\/strong\u003e rack-scale systems\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e36\u003c\/strong\u003e Grace CPUs in GB200 NVL72\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e400 Gb\/s\u003c\/strong\u003e networking support\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003eFY2025 data center revenue was \u003cstrong\u003e$115.2B\u003c\/strong\u003e out of total revenue of \u003cstrong\u003e$130.5B\u003c\/strong\u003e. The calculation is \u003cstrong\u003e$115.2B \/ $130.5B = 88.3%\u003c\/strong\u003e. Q4 FY2025 data center revenue was \u003cstrong\u003e$35.6B\u003c\/strong\u003e out of total revenue of \u003cstrong\u003e$39.3B\u003c\/strong\u003e. The calculation is \u003cstrong\u003e$35.6B \/ $39.3B = 90.6%\u003c\/strong\u003e.\u003c\/p\u003e\n\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Key Resources\u003c\/h2\u003e\n\u003cp\u003e\u003cstrong\u003e$5.42T\u003c\/strong\u003e is the market value anchor for NVIDIA Corporation's resource base. That valuation, combined with a multi-generation chip roadmap and a large developer ecosystem, is a core business asset because it strengthens customer confidence, supplier access, and talent attraction at the same time.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003e\u003cstrong\u003eKey resource\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003eReal-life number or amount\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003eBusiness model effect\u003c\/strong\u003e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eBlackwell\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e208 billion transistors\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eHigh compute density for AI training and inference\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCUDA ecosystem\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003eMore than 4 million developers\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eSwitching costs and software lock-in\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNVIDIA Corporation market value\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$5.42T\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eBrand power, financing strength, and supplier leverage\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGB200 system integration\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e1 Grace CPU + 2 Blackwell GPUs\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003ePlatform-level control over CPU and GPU stack\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eHBM4 and CoWoS access\u003c\/td\u003e\n\u003ctd\u003eHBM4, CoWoS\u003c\/td\u003e\n\u003ctd\u003eSupply access for top-end accelerators\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003eBlackwell\u003c\/strong\u003e, \u003cstrong\u003eRubin\u003c\/strong\u003e, and \u003cstrong\u003eFeynman\u003c\/strong\u003e are the architecture names that carry NVIDIA Corporation's hardware roadmap. Blackwell is the current named architecture with \u003cstrong\u003e208 billion transistors\u003c\/strong\u003e, while Rubin and Feynman extend the roadmap beyond one product cycle. This matters because the architecture sequence is a resource in itself: it keeps customers tied to NVIDIA Corporation's platform planning, not just to one chip sale.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eBlackwell\u003c\/strong\u003e is not only a chip name; it is a design asset that protects compute density, packaging strategy, and system integration. The \u003cstrong\u003e208 billion transistor\u003c\/strong\u003e scale is important because transistor count is one of the clearest physical measures of how much AI work a chip can handle. In business model terms, that resource supports premium pricing, repeat purchases, and long customer planning cycles.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eVera CPU\u003c\/strong\u003e and \u003cstrong\u003eRubin GPU IP\u003c\/strong\u003e are intellectual property resources. The CPU side matters because it keeps the host processor inside the company's platform design, and the GPU side matters because it keeps the AI accelerator inside the same stack. When both layers are owned in-house, NVIDIA Corporation can control system design choices, software compatibility, and roadmap timing more tightly than a chip vendor that depends on outside CPU IP.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eCUDA\u003c\/strong\u003e is the strongest software resource in the canvas. NVIDIA Corporation has said its CUDA ecosystem reaches \u003cstrong\u003emore than 4 million developers\u003c\/strong\u003e. That number matters because developer familiarity becomes switching cost: once engineers write code, tune models, and build workflows around CUDA, moving away from it takes time, retraining, and re-validation. CUDA is not just software; it is the retention layer that protects hardware demand.\u003c\/p\u003e\n\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e2006\u003c\/strong\u003e: CUDA launch year\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eMore than 4 million\u003c\/strong\u003e: CUDA developers\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e208 billion\u003c\/strong\u003e: Blackwell transistor count\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e1\u003c\/strong\u003e Grace CPU in GB200\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e2\u003c\/strong\u003e Blackwell GPUs in GB200\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$5.42T\u003c\/strong\u003e: market value\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eCoWoS\u003c\/strong\u003e wafer access and \u003cstrong\u003eHBM4\u003c\/strong\u003e supply access are physical bottleneck resources. CoWoS advanced packaging is needed to assemble high-end accelerators at scale, and HBM4 is the next-generation high-bandwidth memory class needed for AI workloads that move large data sets quickly. If either one is constrained, chip output is constrained, even when design demand is strong. That makes packaging and memory access part of the resource base, not just the supply chain.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eNVIDIA Corporation\u003c\/strong\u003e brand strength is tied directly to its \u003cstrong\u003e$5.42T\u003c\/strong\u003e market value. That scale gives the company a stronger position with cloud buyers, enterprise customers, foundries, memory suppliers, and packaging partners. In practical terms, the brand does not just signal recognition; it supports demand certainty, bargaining power, and the ability to pull scarce components into the platform.\u003c\/p\u003e\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Value Propositions\u003c\/h2\u003e\n\u003cp\u003eNVIDIA Corporation's value proposition is tied to \u003cstrong\u003e$60.922B\u003c\/strong\u003e FY2024 revenue, \u003cstrong\u003e$47.5B\u003c\/strong\u003e FY2024 data center revenue, up to \u003cstrong\u003e10x\u003c\/strong\u003e lower Mixture-of-experts (MoE) cost per token, up to \u003cstrong\u003e7x\u003c\/strong\u003e faster inference with Dynamo 1.0, and local \u003cstrong\u003e1-trillion-parameter\u003c\/strong\u003e AI on DGX Station.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003cth\u003eValue proposition\u003c\/th\u003e\n\u003cth\u003eFigure 1\u003c\/th\u003e\n\u003cth\u003eFigure 2\u003c\/th\u003e\n\u003cth\u003eCalculation\u003c\/th\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI infrastructure for training and inference\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$60.922B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$47.5B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e77.96%\u003c\/strong\u003e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI infrastructure for training and inference\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$26.044B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$22.563B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e86.64%\u003c\/strong\u003e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eUp to 10x lower MoE cost per token\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e10x\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e10x\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e10x\u003c\/strong\u003e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eUp to 7x faster inference with Dynamo 1.0\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e7x\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e7x\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e7x\u003c\/strong\u003e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLocal 1-trillion-parameter AI on DGX Station\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e1 trillion\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e1 trillion\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e1 trillion\u003c\/strong\u003e\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI factory stack for autonomous workloads\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e72.7%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e78.4%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\n\u003cstrong\u003e5.7\u003c\/strong\u003e percentage points\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cul class=\"lst_crct\"\u003e\n\u003cli\u003e\n\u003cstrong\u003eFY2024 revenue:\u003c\/strong\u003e \u003cstrong\u003e$60.922B\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eFY2024 data center revenue:\u003c\/strong\u003e \u003cstrong\u003e$47.5B\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eFY2024 data center share:\u003c\/strong\u003e \u003cstrong\u003e77.96%\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eQ1 FY2025 revenue:\u003c\/strong\u003e \u003cstrong\u003e$26.044B\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eQ1 FY2025 data center revenue:\u003c\/strong\u003e \u003cstrong\u003e$22.563B\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eQ1 FY2025 data center share:\u003c\/strong\u003e \u003cstrong\u003e86.64%\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eFY2024 GAAP gross margin:\u003c\/strong\u003e \u003cstrong\u003e72.7%\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eQ1 FY2025 GAAP gross margin:\u003c\/strong\u003e \u003cstrong\u003e78.4%\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003eGross margin change:\u003c\/strong\u003e \u003cstrong\u003e5.7\u003c\/strong\u003e percentage points\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eAI infrastructure for training and inference\u003c\/strong\u003e generated \u003cstrong\u003e$47.5B\u003c\/strong\u003e of \u003cstrong\u003e$60.922B\u003c\/strong\u003e in FY2024, or \u003cstrong\u003e77.96%\u003c\/strong\u003e. Q1 FY2025 data center revenue was \u003cstrong\u003e$22.563B\u003c\/strong\u003e of \u003cstrong\u003e$26.044B\u003c\/strong\u003e, or \u003cstrong\u003e86.64%\u003c\/strong\u003e.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eUp to 10x lower MoE cost per token\u003c\/strong\u003e and \u003cstrong\u003eup to 7x faster inference with Dynamo 1.0\u003c\/strong\u003e are the clearest unit-economics numbers in the stack: \u003cstrong\u003e10x\u003c\/strong\u003e lower cost per token and \u003cstrong\u003e7x\u003c\/strong\u003e faster inference.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eLocal 1-trillion-parameter AI on DGX Station\u003c\/strong\u003e places \u003cstrong\u003e1 trillion\u003c\/strong\u003e parameters in a local setup.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eAI factory stack for autonomous workloads\u003c\/strong\u003e connects to gross margin of \u003cstrong\u003e72.7%\u003c\/strong\u003e in FY2024 and \u003cstrong\u003e78.4%\u003c\/strong\u003e in Q1 FY2025.\u003c\/p\u003e\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Customer Relationships\u003c\/h2\u003e\n\n\u003cp\u003eFY2025 revenue was \u003cstrong\u003e$130.5 billion\u003c\/strong\u003e. Data center revenue was \u003cstrong\u003e$115.2 billion\u003c\/strong\u003e. Fiscal year ended January 26, 2025.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003eCustomer relationship channel\u003c\/td\u003e\n\u003ctd\u003eReal-life numbers, dates, and named examples\u003c\/td\u003e\n\u003ctd\u003eBusiness model effect\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eStrategic enterprise and cloud partnerships\u003c\/td\u003e\n\u003ctd\u003e\n\u003cstrong\u003e$130.5 billion\u003c\/strong\u003e; \u003cstrong\u003e$115.2 billion\u003c\/strong\u003e; Microsoft Azure; Google Cloud; AWS; Oracle Cloud Infrastructure; CoreWeave\u003c\/td\u003e\n\u003ctd\u003eLarge cloud buyers place repeated orders and expand deployed capacity\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCo-development with frontier AI labs\u003c\/td\u003e\n\u003ctd\u003eOpenAI; Anthropic; Meta; xAI; Mistral AI; Cohere; Blackwell B200; \u003cstrong\u003e208 billion\u003c\/strong\u003e transistors; GB200 NVL72; \u003cstrong\u003e72\u003c\/strong\u003e GPUs; \u003cstrong\u003e36\u003c\/strong\u003e Grace CPUs\u003c\/td\u003e\n\u003ctd\u003eJoint engineering locks customers into training and inference planning cycles\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLong-term platform roadmap support\u003c\/td\u003e\n\u003ctd\u003eHopper; Blackwell; CUDA; TensorRT; NVIDIA AI Enterprise; 2024; January 26, 2025\u003c\/td\u003e\n\u003ctd\u003eCustomers plan multiyear refreshes around software and hardware continuity\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDirect sales to large enterprises\u003c\/td\u003e\n\u003ctd\u003eNVIDIA AI Enterprise; DGX systems; \u003cstrong\u003e$115.2 billion\u003c\/strong\u003e data center revenue\u003c\/td\u003e\n\u003ctd\u003eDirect account control supports software, systems, and support sales\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOEM enablement for server shipping\u003c\/td\u003e\n\u003ctd\u003eDell Technologies; HPE; Lenovo; Supermicro; Cisco; GB200 NVL72; \u003cstrong\u003e72\u003c\/strong\u003e GPUs; \u003cstrong\u003e36\u003c\/strong\u003e Grace CPUs\u003c\/td\u003e\n\u003ctd\u003eOEMs ship validated servers faster when the platform spec is fixed\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003eStrategic enterprise and cloud partnerships.\u003c\/strong\u003e \u003cstrong\u003e$130.5 billion\u003c\/strong\u003e; \u003cstrong\u003e$115.2 billion\u003c\/strong\u003e; Microsoft Azure; Google Cloud; AWS; Oracle Cloud Infrastructure; CoreWeave.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003eMicrosoft Azure\u003c\/li\u003e\n\u003cli\u003eGoogle Cloud\u003c\/li\u003e\n\u003cli\u003eAWS\u003c\/li\u003e\n\u003cli\u003eOracle Cloud Infrastructure\u003c\/li\u003e\n\u003cli\u003eCoreWeave\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eCo-development with frontier AI labs.\u003c\/strong\u003e Blackwell B200: \u003cstrong\u003e208 billion\u003c\/strong\u003e transistors. GB200 NVL72: \u003cstrong\u003e72\u003c\/strong\u003e GPUs and \u003cstrong\u003e36\u003c\/strong\u003e Grace CPUs.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003eOpenAI\u003c\/li\u003e\n\u003cli\u003eAnthropic\u003c\/li\u003e\n\u003cli\u003eMeta\u003c\/li\u003e\n\u003cli\u003exAI\u003c\/li\u003e\n\u003cli\u003eMistral AI\u003c\/li\u003e\n\u003cli\u003eCohere\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eLong-term platform roadmap support.\u003c\/strong\u003e Hopper; Blackwell; CUDA; TensorRT; NVIDIA AI Enterprise; 2024; January 26, 2025.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003eHopper\u003c\/li\u003e\n\u003cli\u003eBlackwell\u003c\/li\u003e\n\u003cli\u003eCUDA\u003c\/li\u003e\n\u003cli\u003eTensorRT\u003c\/li\u003e\n\u003cli\u003eNVIDIA AI Enterprise\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eDirect sales to large enterprises.\u003c\/strong\u003e NVIDIA AI Enterprise; DGX systems; \u003cstrong\u003e$115.2 billion\u003c\/strong\u003e data center revenue.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003eNVIDIA AI Enterprise\u003c\/li\u003e\n\u003cli\u003eDGX systems\u003c\/li\u003e\n\u003cli\u003eDirect enterprise accounts\u003c\/li\u003e\n\u003cli\u003eData center customers\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eOEM enablement for server shipping.\u003c\/strong\u003e Dell Technologies; HPE; Lenovo; Supermicro; Cisco; GB200 NVL72; \u003cstrong\u003e72\u003c\/strong\u003e GPUs; \u003cstrong\u003e36\u003c\/strong\u003e Grace CPUs.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003eDell Technologies\u003c\/li\u003e\n\u003cli\u003eHPE\u003c\/li\u003e\n\u003cli\u003eLenovo\u003c\/li\u003e\n\u003cli\u003eSupermicro\u003c\/li\u003e\n\u003cli\u003eCisco\u003c\/li\u003e\n\u003c\/ul\u003e\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Channels\u003c\/h2\u003e\n\u003cp\u003eNVIDIA Corporation's channels center on \u003cstrong\u003e8-GPU\u003c\/strong\u003e cloud instances, \u003cstrong\u003e8-GPU\u003c\/strong\u003e OEM servers, \u003cstrong\u003e8-GPU\u003c\/strong\u003e direct systems, \u003cstrong\u003e72-GPU\u003c\/strong\u003e rack-scale systems, and accelerator memory bands from \u003cstrong\u003e24GB\u003c\/strong\u003e to \u003cstrong\u003e192GB\u003c\/strong\u003e.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eHyperscaler cloud instances:\u003c\/strong\u003e AWS EC2 P5, Azure ND H100 v5, Google Cloud A3, and Oracle Cloud BM.GPU.H100.8 each use \u003cstrong\u003e8 H100 GPUs\u003c\/strong\u003e.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eOEM server partners:\u003c\/strong\u003e Dell PowerEdge XE9680 and Supermicro SYS-821GE-TNHR are \u003cstrong\u003e8-GPU\u003c\/strong\u003e server classes.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eDirect enterprise sales:\u003c\/strong\u003e DGX H100 uses \u003cstrong\u003e8 H100 GPUs\u003c\/strong\u003e and \u003cstrong\u003e640GB\u003c\/strong\u003e total GPU memory; DGX B200 uses \u003cstrong\u003e8 B200 GPUs\u003c\/strong\u003e and \u003cstrong\u003e1,536GB\u003c\/strong\u003e total GPU memory.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eNVIDIA-branded AI systems:\u003c\/strong\u003e DGX GH200 uses \u003cstrong\u003e256 GH200 superchips\u003c\/strong\u003e and \u003cstrong\u003e144TB\u003c\/strong\u003e shared memory; GB200 NVL72 uses \u003cstrong\u003e72 GPUs\u003c\/strong\u003e and \u003cstrong\u003e36 CPUs\u003c\/strong\u003e.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003eChannel\u003c\/td\u003e\n\u003ctd\u003eReal-life numeric configuration\u003c\/td\u003e\n\u003ctd\u003eNamed example\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eHyperscaler cloud instances\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e8 H100 GPUs\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eAWS EC2 P5, Azure ND H100 v5, Google Cloud A3, Oracle Cloud BM.GPU.H100.8\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOEM server partners\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e8 GPUs\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eDell PowerEdge XE9680, Supermicro SYS-821GE-TNHR\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eDirect enterprise sales\u003c\/td\u003e\n\u003ctd\u003e\n\u003cstrong\u003e8 H100 GPUs\u003c\/strong\u003e, \u003cstrong\u003e640GB\u003c\/strong\u003e; \u003cstrong\u003e8 B200 GPUs\u003c\/strong\u003e, \u003cstrong\u003e1,536GB\u003c\/strong\u003e\n\u003c\/td\u003e\n\u003ctd\u003eDGX H100, DGX B200\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eNVIDIA-branded AI systems\u003c\/td\u003e\n\u003ctd\u003e\n\u003cstrong\u003e256\u003c\/strong\u003e, \u003cstrong\u003e144TB\u003c\/strong\u003e; \u003cstrong\u003e72 GPUs\u003c\/strong\u003e, \u003cstrong\u003e36 CPUs\u003c\/strong\u003e\n\u003c\/td\u003e\n\u003ctd\u003eDGX GH200, GB200 NVL72\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003eData center and edge platform releases:\u003c\/strong\u003e H100 uses \u003cstrong\u003e80GB\u003c\/strong\u003e HBM3; H200 uses \u003cstrong\u003e141GB\u003c\/strong\u003e HBM3e; B200 uses \u003cstrong\u003e192GB\u003c\/strong\u003e HBM3e; L4 uses \u003cstrong\u003e24GB\u003c\/strong\u003e; L40S uses \u003cstrong\u003e48GB\u003c\/strong\u003e; Jetson AGX Orin uses \u003cstrong\u003e64GB\u003c\/strong\u003e.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003ePlatform\u003c\/td\u003e\n\u003ctd\u003eMemory\u003c\/td\u003e\n\u003ctd\u003eCategory\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eH100\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e80GB\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eData center\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eH200\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e141GB\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eData center\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eB200\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e192GB\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eData center\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eL4\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e24GB\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eEdge\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eL40S\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e48GB\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eData center and edge\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eJetson AGX Orin\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e64GB\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eEdge\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e8\u003c\/strong\u003e GPUs per cloud instance class\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e8\u003c\/strong\u003e GPUs per DGX H100 system\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e8\u003c\/strong\u003e GPUs per DGX B200 system\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e72\u003c\/strong\u003e GPUs and \u003cstrong\u003e36\u003c\/strong\u003e CPUs per GB200 NVL72 system\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e256\u003c\/strong\u003e GH200 superchips and \u003cstrong\u003e144TB\u003c\/strong\u003e shared memory per DGX GH200 system\u003c\/li\u003e\n\u003c\/ul\u003e\n\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Customer Segments\u003c\/h2\u003e\n\u003cp\u003e\u003cstrong\u003e$60.9B\u003c\/strong\u003e total FY2024 revenue was concentrated in Data Center at \u003cstrong\u003e$47.5B\u003c\/strong\u003e, or \u003cstrong\u003e78.0%\u003c\/strong\u003e of total revenue. Gaming was \u003cstrong\u003e$10.4B\u003c\/strong\u003e, Pro Visualization was \u003cstrong\u003e$1.6B\u003c\/strong\u003e, Automotive was \u003cstrong\u003e$1.1B\u003c\/strong\u003e, and OEM and other was \u003cstrong\u003e$0.4B\u003c\/strong\u003e.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003eCustomer segment\u003c\/td\u003e\n\u003ctd\u003ePublic revenue line\u003c\/td\u003e\n\u003ctd\u003eFY2024 amount\u003c\/td\u003e\n\u003ctd\u003eShare of $60.9B\u003c\/td\u003e\n\u003ctd\u003eCustomer role\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eHyperscalers and cloud providers\u003c\/td\u003e\n\u003ctd\u003eData Center\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$47.5B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e78.0%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eLarge-scale GPU and networking buys\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eFrontier AI labs\u003c\/td\u003e\n\u003ctd\u003eData Center\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$47.5B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e78.0%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eModel training and inference clusters\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eLarge enterprises\u003c\/td\u003e\n\u003ctd\u003eData Center\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$47.5B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e78.0%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003ePrivate AI infrastructure and software\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOEM server buyers\u003c\/td\u003e\n\u003ctd\u003eOEM and other\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$0.4B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e0.7%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eServer OEM and system builder demand\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI PC and workstation users\u003c\/td\u003e\n\u003ctd\u003eGaming and Pro Visualization\u003c\/td\u003e\n\u003ctd\u003e\n\u003cstrong\u003e$10.4B\u003c\/strong\u003e and \u003cstrong\u003e$1.6B\u003c\/strong\u003e\n\u003c\/td\u003e\n\u003ctd\u003e\n\u003cstrong\u003e17.1%\u003c\/strong\u003e and \u003cstrong\u003e2.6%\u003c\/strong\u003e\n\u003c\/td\u003e\n\u003ctd\u003ePC, creator, and workstation demand\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003eHyperscalers and cloud providers\u003c\/strong\u003e are the largest customer segment by revenue. Data Center revenue of \u003cstrong\u003e$47.5B\u003c\/strong\u003e was \u003cstrong\u003e217%\u003c\/strong\u003e higher year over year and represented \u003cstrong\u003e78.0%\u003c\/strong\u003e of NVIDIA's \u003cstrong\u003e$60.9B\u003c\/strong\u003e total revenue. This segment includes the largest infrastructure buyers that fund multi-billion-dollar capital spending on servers, accelerators, and networking.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e$47.5B\u003c\/strong\u003e Data Center revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e78.0%\u003c\/strong\u003e of total revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e217%\u003c\/strong\u003e year-over-year growth\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eFrontier AI labs\u003c\/strong\u003e sit inside Data Center revenue and do not have a separate public revenue line. Their purchases are part of the \u003cstrong\u003e$47.5B\u003c\/strong\u003e Data Center total, which means their demand is embedded in the same segment as cloud operators and other large infrastructure buyers.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003eNo separate public revenue line\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003eIncluded in \u003cstrong\u003e$47.5B\u003c\/strong\u003e Data Center revenue\u003c\/li\u003e\n\u003cli\u003eIncluded in \u003cstrong\u003e78.0%\u003c\/strong\u003e of total revenue\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eLarge enterprises\u003c\/strong\u003e are also embedded in Data Center revenue. They do not have a separate disclosed revenue line, so their contribution is part of the same \u003cstrong\u003e$47.5B\u003c\/strong\u003e total. This matters because enterprise AI buying is visible in the Data Center number, not in a standalone enterprise segment.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003eNo separate public revenue line\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003eIncluded in \u003cstrong\u003e$47.5B\u003c\/strong\u003e Data Center revenue\u003c\/li\u003e\n\u003cli\u003eIncluded in \u003cstrong\u003e78.0%\u003c\/strong\u003e of total revenue\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eOEM server buyers\u003c\/strong\u003e are the smallest clearly disclosed B2B hardware segment in NVIDIA's reporting. OEM and other revenue was \u003cstrong\u003e$0.4B\u003c\/strong\u003e, or \u003cstrong\u003e0.7%\u003c\/strong\u003e of total revenue. This segment captures server OEM and system-builder demand rather than direct cloud purchases.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e$0.4B\u003c\/strong\u003e OEM and other revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e0.7%\u003c\/strong\u003e of total revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e1\u003c\/strong\u003e public reporting line for this segment\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eAI PC and workstation users\u003c\/strong\u003e map mainly to Gaming and Pro Visualization. Gaming revenue was \u003cstrong\u003e$10.4B\u003c\/strong\u003e, or \u003cstrong\u003e17.1%\u003c\/strong\u003e of total revenue, and Pro Visualization revenue was \u003cstrong\u003e$1.6B\u003c\/strong\u003e, or \u003cstrong\u003e2.6%\u003c\/strong\u003e of total revenue. Together, those two segments represented \u003cstrong\u003e19.7%\u003c\/strong\u003e of FY2024 revenue.\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e$10.4B\u003c\/strong\u003e Gaming revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$1.6B\u003c\/strong\u003e Pro Visualization revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e19.7%\u003c\/strong\u003e combined share of total revenue\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003eSegment\u003c\/td\u003e\n\u003ctd\u003eFY2024 revenue\u003c\/td\u003e\n\u003ctd\u003eShare of $60.9B\u003c\/td\u003e\n\u003ctd\u003eCustomer mix signal\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eData Center\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$47.5B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e78.0%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eHyperscalers, frontier AI labs, large enterprises\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eGaming\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$10.4B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e17.1%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eAI PC and consumer GPU users\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003ePro Visualization\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$1.6B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e2.6%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eWorkstation and creator users\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAutomotive\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$1.1B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e1.8%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eVehicle and mobility customers\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eOEM and other\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e$0.4B\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003e\u003cstrong\u003e0.7%\u003c\/strong\u003e\u003c\/td\u003e\n\u003ctd\u003eServer OEM and system builders\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Cost Structure\u003c\/h2\u003e\n\u003cp\u003e\u003cstrong\u003e$130.497B\u003c\/strong\u003e revenue, \u003cstrong\u003e$32.624B\u003c\/strong\u003e cost of revenue, \u003cstrong\u003e75.0%\u003c\/strong\u003e gross margin, \u003cstrong\u003e$12.914B\u003c\/strong\u003e R\u0026amp;D, and \u003cstrong\u003e$5.5B\u003c\/strong\u003e export-control-related charge.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003cth\u003eFiscal year ended\u003c\/th\u003e\n\u003cth\u003eRevenue\u003c\/th\u003e\n\u003cth\u003eGross margin\u003c\/th\u003e\n\u003cth\u003eCost of revenue\u003c\/th\u003e\n\u003cth\u003eGross profit\u003c\/th\u003e\n\u003cth\u003eR\u0026amp;D\u003c\/th\u003e\n\u003cth\u003eR\u0026amp;D \/ revenue\u003c\/th\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eJanuary 29, 2023\u003c\/td\u003e\n\u003ctd\u003e$26.974B\u003c\/td\u003e\n\u003ctd\u003e56.9%\u003c\/td\u003e\n\u003ctd\u003e$11.626B\u003c\/td\u003e\n\u003ctd\u003e$15.348B\u003c\/td\u003e\n\u003ctd\u003e$7.339B\u003c\/td\u003e\n\u003ctd\u003e27.2%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eJanuary 28, 2024\u003c\/td\u003e\n\u003ctd\u003e$60.922B\u003c\/td\u003e\n\u003ctd\u003e72.7%\u003c\/td\u003e\n\u003ctd\u003e$16.632B\u003c\/td\u003e\n\u003ctd\u003e$44.290B\u003c\/td\u003e\n\u003ctd\u003e$8.687B\u003c\/td\u003e\n\u003ctd\u003e14.3%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eJanuary 26, 2025\u003c\/td\u003e\n\u003ctd\u003e$130.497B\u003c\/td\u003e\n\u003ctd\u003e75.0%\u003c\/td\u003e\n\u003ctd\u003e$32.624B\u003c\/td\u003e\n\u003ctd\u003e$97.873B\u003c\/td\u003e\n\u003ctd\u003e$12.914B\u003c\/td\u003e\n\u003ctd\u003e9.9%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003eR\u0026amp;D for chips, CPUs, and software\u003c\/strong\u003e\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e$7.339B\u003c\/strong\u003e in FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$8.687B\u003c\/strong\u003e in FY2024\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$12.914B\u003c\/strong\u003e in FY2025\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e27.2%\u003c\/strong\u003e of revenue in FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e14.3%\u003c\/strong\u003e of revenue in FY2024\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e9.9%\u003c\/strong\u003e of revenue in FY2025\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e18.4%\u003c\/strong\u003e FY2024 growth versus FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e48.7%\u003c\/strong\u003e FY2025 growth versus FY2024\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eAdvanced packaging and memory procurement\u003c\/strong\u003e\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e$11.626B\u003c\/strong\u003e cost of revenue in FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$16.632B\u003c\/strong\u003e cost of revenue in FY2024\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$32.624B\u003c\/strong\u003e cost of revenue in FY2025\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e43.1%\u003c\/strong\u003e cost of revenue \/ revenue in FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e27.3%\u003c\/strong\u003e cost of revenue \/ revenue in FY2024\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e25.0%\u003c\/strong\u003e cost of revenue \/ revenue in FY2025\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e56.9%\u003c\/strong\u003e gross margin in FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e72.7%\u003c\/strong\u003e gross margin in FY2024\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e75.0%\u003c\/strong\u003e gross margin in FY2025\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eInventory charges and write-downs\u003c\/strong\u003e\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003eQ1 FY2026 ended April 27, 2025: \u003cstrong\u003e$5.5B\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003cth\u003ePeriod\u003c\/th\u003e\n\u003cth\u003eCharge\u003c\/th\u003e\n\u003cth\u003eCost area\u003c\/th\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eQ1 FY2026 ended April 27, 2025\u003c\/td\u003e\n\u003ctd\u003e$5.5B\u003c\/td\u003e\n\u003ctd\u003eH20 export restriction\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003eSupply chain and capacity constraints\u003c\/strong\u003e\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cstrong\u003e$26.974B\u003c\/strong\u003e revenue in FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$60.922B\u003c\/strong\u003e revenue in FY2024\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$130.497B\u003c\/strong\u003e revenue in FY2025\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e125.9%\u003c\/strong\u003e revenue growth in FY2024 versus FY2023\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e114.3%\u003c\/strong\u003e revenue growth in FY2025 versus FY2024\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$11.626B\u003c\/strong\u003e, \u003cstrong\u003e$16.632B\u003c\/strong\u003e, and \u003cstrong\u003e$32.624B\u003c\/strong\u003e cost of revenue across FY2023, FY2024, and FY2025\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e43.1%\u003c\/strong\u003e, \u003cstrong\u003e27.3%\u003c\/strong\u003e, and \u003cstrong\u003e25.0%\u003c\/strong\u003e cost of revenue \/ revenue across FY2023, FY2024, and FY2025\u003c\/li\u003e\n\u003c\/ul\u003e\n\n\u003cp\u003e\u003cstrong\u003eCompliance and export-control costs\u003c\/strong\u003e\u003c\/p\u003e\n\u003cul\u003e\n\u003cli\u003eQ1 FY2026 ended April 27, 2025: \u003cstrong\u003e$5.5B\u003c\/strong\u003e\n\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003e2025\u003c\/strong\u003e\u003c\/li\u003e\n\u003cli\u003e\u003cstrong\u003e2026\u003c\/strong\u003e\u003c\/li\u003e\n\u003c\/ul\u003e\u003ch2\u003eNVIDIA Corporation - Canvas Business Model: Revenue Streams\u003c\/h2\u003e\n\n\u003cp\u003e\u003cstrong\u003e$130.5b\u003c\/strong\u003e FY2025 revenue; \u003cstrong\u003e$39.3b\u003c\/strong\u003e Q4 FY2025 revenue.\u003c\/p\u003e\n\n\u003ctable\u003e\n\u003ctr\u003e\n\u003ctd\u003eRevenue stream\u003c\/td\u003e\n\u003ctd\u003eFY2025 revenue\u003c\/td\u003e\n\u003ctd\u003eFY2025 share of $130.5b\u003c\/td\u003e\n\u003ctd\u003eQ4 FY2025 revenue\u003c\/td\u003e\n\u003ctd\u003eQ4 share of $39.3b\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eData Center platforms and systems\u003c\/td\u003e\n\u003ctd\u003e$115.2b\u003c\/td\u003e\n\u003ctd\u003e88.3%\u003c\/td\u003e\n\u003ctd\u003e$35.6b\u003c\/td\u003e\n\u003ctd\u003e90.6%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eEdge Computing products\u003c\/td\u003e\n\u003ctd\u003e$1.7b\u003c\/td\u003e\n\u003ctd\u003e1.3%\u003c\/td\u003e\n\u003ctd\u003e$0.570b\u003c\/td\u003e\n\u003ctd\u003e1.5%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eAI hardware sales for inference and training\u003c\/td\u003e\n\u003ctd\u003e$115.2b\u003c\/td\u003e\n\u003ctd\u003e88.3%\u003c\/td\u003e\n\u003ctd\u003e$11.0b Blackwell\u003c\/td\u003e\n\u003ctd\u003e28.0%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eEnterprise AI and AI PC systems\u003c\/td\u003e\n\u003ctd\u003e$13.3b\u003c\/td\u003e\n\u003ctd\u003e10.2%\u003c\/td\u003e\n\u003ctd\u003e$3.0b\u003c\/td\u003e\n\u003ctd\u003e7.6%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003ctr\u003e\n\u003ctd\u003eCloud and OEM platform deployments\u003c\/td\u003e\n\u003ctd\u003e$115.5b\u003c\/td\u003e\n\u003ctd\u003e88.5%\u003c\/td\u003e\n\u003ctd\u003e$35.6b Data Center; $0.090b OEM and Other\u003c\/td\u003e\n\u003ctd\u003e90.6%; 0.2%\u003c\/td\u003e\n\u003c\/tr\u003e\n\u003c\/table\u003e\n\n\u003cp\u003e\u003cstrong\u003e$115.2b\u003c\/strong\u003e Data Center revenue; \u003cstrong\u003e$11.0b\u003c\/strong\u003e Blackwell revenue in Q4 FY2025; \u003cstrong\u003e$11.0b \/ $35.6b = 30.9%\u003c\/strong\u003e; \u003cstrong\u003e$11.0b \/ $39.3b = 28.0%\u003c\/strong\u003e.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eData Center platforms and systems\u003c\/strong\u003e: \u003cstrong\u003e$22.6b\u003c\/strong\u003e Q1 FY2025; \u003cstrong\u003e$26.3b\u003c\/strong\u003e Q2 FY2025; \u003cstrong\u003e$30.8b\u003c\/strong\u003e Q3 FY2025; \u003cstrong\u003e$35.6b\u003c\/strong\u003e Q4 FY2025.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eEdge Computing products\u003c\/strong\u003e: \u003cstrong\u003e$1.7b\u003c\/strong\u003e Automotive FY2025; \u003cstrong\u003e$0.570b\u003c\/strong\u003e Q4 FY2025; \u003cstrong\u003e1.3%\u003c\/strong\u003e of FY2025 revenue.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eAI hardware sales for inference and training\u003c\/strong\u003e: \u003cstrong\u003e$115.2b\u003c\/strong\u003e Data Center FY2025; \u003cstrong\u003e$11.0b\u003c\/strong\u003e Blackwell in Q4 FY2025; \u003cstrong\u003e88.3%\u003c\/strong\u003e of FY2025 revenue from Data Center.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eEnterprise AI and AI PC systems\u003c\/strong\u003e: \u003cstrong\u003e$11.4b\u003c\/strong\u003e Gaming FY2025; \u003cstrong\u003e$1.9b\u003c\/strong\u003e Professional Visualization FY2025; \u003cstrong\u003e$13.3b\u003c\/strong\u003e combined; \u003cstrong\u003e10.2%\u003c\/strong\u003e of FY2025 revenue.\u003c\/p\u003e\n\n\u003cp\u003e\u003cstrong\u003eCloud and OEM platform deployments\u003c\/strong\u003e: \u003cstrong\u003e$115.2b\u003c\/strong\u003e Data Center FY2025; \u003cstrong\u003e$35.6b\u003c\/strong\u003e Data Center Q4 FY2025; \u003cstrong\u003e$0.3b\u003c\/strong\u003e OEM and Other FY2025; \u003cstrong\u003e0.2%\u003c\/strong\u003e of FY2025 revenue.\u003c\/p\u003e\n\n\u003cul class=\"lst_crct\"\u003e\n\u003cli\u003e\n\u003cstrong\u003e$26.0b\u003c\/strong\u003e Q1 FY2025 total revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$30.0b\u003c\/strong\u003e Q2 FY2025 total revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$35.1b\u003c\/strong\u003e Q3 FY2025 total revenue\u003c\/li\u003e\n\u003cli\u003e\n\u003cstrong\u003e$39.3b\u003c\/strong\u003e Q4 FY2025 total revenue\u003c\/li\u003e\n\u003c\/ul\u003e","brand":"dcf.fm","offers":[{"title":"Default Title","offer_id":44601615188117,"sku":"nvda-business-model-canvas","price":7.0,"currency_code":"USD","in_stock":true}],"thumbnail_url":"\/\/cdn.shopify.com\/s\/files\/1\/0630\/5189\/0837\/files\/nvda-business-model-canvas.png?v=1740200917","url":"https:\/\/dcf-analysis.com\/products\/nvda-business-model-canvas","provider":"AI-Powered Discounted Cash Flow Model Templates","version":"1.0","type":"link"}