{
  "data_version": "1",
  "schema": "https://whatledto.com/data",
  "slug": "nvidia-ai-compute-origins",
  "title": "How Nvidia Came to Own AI Compute: From CUDA to the H100 Shortage",
  "url": "https://whatledto.com/events/nvidia-ai-compute-origins",
  "markdown": "https://whatledto.com/events/nvidia-ai-compute-origins.md",
  "feed": "https://whatledto.com/events/nvidia-ai-compute-origins/feed.xml",
  "vertical": "tech-industry",
  "topic": "Tech industry",
  "status": "live",
  "tags": [
    {
      "slug": "ai-accelerators",
      "name": "AI accelerators",
      "url": "https://whatledto.com/tags/ai-accelerators"
    },
    {
      "slug": "data-center-gpus",
      "name": "Data center GPUs",
      "url": "https://whatledto.com/tags/data-center-gpus"
    },
    {
      "slug": "nvidia",
      "name": "Nvidia",
      "url": "https://whatledto.com/tags/nvidia"
    }
  ],
  "created_at": "2026-09-14T03:30:20.276349+00:00",
  "updated_at": "2026-10-09T09:16:18.088281+00:00",
  "last_researched_at": null,
  "tldr": "Nvidia’s AI-computing position began with CUDA in 2006, which made GPUs programmable for general-purpose parallel computing. AlexNet’s 2012 ImageNet result showed that a deep neural network could be trained effectively with GPUs, and Nvidia later built Tensor Cores and data-center products such as the V100 and A100 around AI workloads. ChatGPT’s November 2022 release coincided with a broader rush for generative-AI computing capacity; Nvidia said in May 2023 that it was increasing supply to meet surging demand. By fiscal 2024, Nvidia reported $47.5 billion in data-center revenue, up 217% year over year, but the cited results do not separate H100 sales from other data-center products or quantify the shortage directly.",
  "hero_image": "https://images.unsplash.com/photo-1697952431907-8542919a16b3?ixid=M3w5OTg1Nzd8MHwxfGFsbHx8fHx8fHx8fDE3ODk3MjIzMzV8&ixlib=rb-4.1.0&w=1600&q=80&auto=format&fit=crop",
  "hero_credit": {
    "source": "unsplash",
    "photo_id": "b_GcLCaKt94",
    "photographer": "Albert Stoynov",
    "photographer_url": "https://unsplash.com/@albertstoynov?utm_source=whatledto&utm_medium=referral",
    "photo_url": "https://unsplash.com/photos/a-black-and-white-photo-of-a-computer-motherboard-b_GcLCaKt94?utm_source=whatledto&utm_medium=referral",
    "description": "PCB of a Graphics Card.",
    "license": "",
    "license_url": ""
  },
  "entities": [
    {
      "id": "nvidia",
      "url": "https://whatledto.com/who/nvidia",
      "json": "https://whatledto.com/who/nvidia.json",
      "name": "Nvidia",
      "kind": "company",
      "role": "Built the CUDA software platform and successive data-center GPU generations that supplied the AI-computing demand described in the timeline."
    },
    {
      "id": "openai",
      "url": "https://whatledto.com/who/openai",
      "json": "https://whatledto.com/who/openai.json",
      "name": "OpenAI",
      "kind": "company",
      "role": "Released ChatGPT as a research preview in November 2022."
    },
    {
      "id": "jensen-huang",
      "url": "https://whatledto.com/who/jensen-huang",
      "json": "https://whatledto.com/who/jensen-huang.json",
      "name": "Jensen Huang",
      "kind": "person",
      "role": "Nvidia’s CEO, who said in May 2023 that the company was significantly increasing supply to meet surging demand."
    },
    {
      "id": "alexnet",
      "name": "AlexNet",
      "kind": "product",
      "role": "The 2012 convolutional neural network whose ImageNet result demonstrated the practical value of GPU-based deep-learning training."
    },
    {
      "id": "ampere-a100",
      "name": "Ampere A100",
      "kind": "product",
      "role": "Nvidia’s 2020 data-center GPU, shipped in volume for AI training and inference and adopted by major cloud providers."
    },
    {
      "id": "cuda",
      "name": "CUDA",
      "kind": "product",
      "role": "Nvidia’s GPU-computing architecture and C development environment, introduced in 2006 to make GPUs usable for general-purpose parallel workloads."
    },
    {
      "id": "chatgpt",
      "name": "ChatGPT",
      "kind": "product",
      "role": "OpenAI’s conversational model, whose release became a visible trigger for broader generative-AI demand."
    },
    {
      "id": "h100",
      "name": "H100",
      "kind": "product",
      "role": "Nvidia’s later data-center GPU generation; fiscal-2024 reporting linked H100-system shipments to the sharp increase in data-center revenue."
    },
    {
      "id": "volta-v100",
      "name": "Volta V100",
      "kind": "product",
      "role": "Nvidia’s 2017 data-center GPU platform, which introduced Tensor Cores for deep-learning matrix calculations."
    }
  ],
  "nodes": [
    {
      "id": 1222,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1222",
      "date": "2006-11-08",
      "date_precision": "day",
      "title": "Nvidia unveils CUDA: the GPU becomes a general-purpose processor with a C compiler",
      "summary": "Nvidia introduced CUDA as a new architecture for computing on its graphics processors and the first C-compiler development environment for the GPU, pitching hundreds of on-chip cores working in parallel at up to 100 times the speed of conventional approaches. It is the software layer that later let researchers train neural networks on gaming cards, and the ecosystem rivals still have to work around.",
      "type": "launch",
      "entities": [
        "Nvidia"
      ],
      "parties": [
        "nvidia"
      ],
      "published_at": "2026-09-14T03:30:20.278949+00:00",
      "note": "",
      "image": "",
      "image_credit": {
        "credit": "",
        "source_url": "",
        "image_url": "",
        "terms_url": "",
        "mode": "licensed",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://web.archive.org/web/20070105114450/http://www.nvidia.com/object/IO_37226.html",
          "outlet": "NVIDIA press release (Wayback Machine)",
          "published": "2006-11-08",
          "quote": "NVIDIA Corporation (Nasdaq: NVDA), the worldwide leader in graphics processors, today unveiled NVIDIA CUDA technology, a fundamentally new architecture for computing on NVIDIA graphics processing units (GPUs), and the industry's first C-compiler development environment for the GPU."
        }
      ]
    },
    {
      "id": 1223,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1223",
      "date": "2012-12-03",
      "date_precision": "day",
      "title": "AlexNet: a deep network trained on two GPUs wins ImageNet by a wide margin",
      "summary": "Krizhevsky, Sutskever and Hinton reported a 60-million-parameter convolutional network that cut ImageNet's top-5 error to 18.9 percent, far below the previous state of the art, and credited \"a very efficient GPU implementation\" for making the training feasible. The result is the moment deep learning and Nvidia GPUs became tied together in research practice.",
      "type": "report",
      "entities": [
        "Nvidia",
        "Geoffrey Hinton",
        "Ilya Sutskever"
      ],
      "parties": [
        "nvidia",
        "geoffrey-hinton",
        "ilya-sutskever"
      ],
      "published_at": "2026-09-14T03:30:20.279050+00:00",
      "note": "",
      "image": "",
      "image_credit": {
        "credit": "",
        "source_url": "",
        "image_url": "",
        "terms_url": "",
        "mode": "licensed",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://proceedings.neurips.cc/paper/2012/hash/c399862d3b9d6b76c8436e924a68c45b-Abstract.html",
          "outlet": "NeurIPS 2012 proceedings",
          "published": "2012-12-03",
          "quote": "On the test data, we achieved top-1 and top-5 error rates of 39.7% and 18.9% which is considerably better than the previous state-of-the-art results. The neural network, which has 60 million parameters and 500,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and two globally connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of convolutional nets."
        }
      ]
    },
    {
      "id": 1224,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1224",
      "date": "2017-05-10",
      "date_precision": "day",
      "title": "Volta V100 adds Tensor Cores: the first data-center GPU designed around AI training",
      "summary": "At GTC 2017 Nvidia launched the Volta platform and the Tesla V100, with 640 Tensor Cores delivering 120 teraflops of deep-learning performance, which it equated to 100 CPUs. Tensor Cores made matrix math a first-class hardware feature, the design choice every later generation has extended.",
      "type": "launch",
      "entities": [
        "Nvidia",
        "Jensen Huang"
      ],
      "parties": [
        "nvidia",
        "jensen-huang"
      ],
      "published_at": "2026-09-14T03:30:20.279097+00:00",
      "note": "",
      "image": "/images/nodes/nvidia-ai-compute-origins/1224-51903704.jpg",
      "image_credit": {
        "credit": "NVIDIA, press image",
        "source_url": "https://nvidianews.nvidia.com/news/nvidia-launches-revolutionary-volta-gpu-platform-fueling-next-era-of-ai-and-high-performance-computing",
        "image_url": "https://iprsoftwaremedia.com/219/files/20174/5913501aa13835537f3aee05_NVIDIA-Telsa-V100/NVIDIA-Telsa-V100_a160e73a-2bd1-4a26-9a2f-9c4e34bd9fcb-prv.jpg?v=a160e73a-2bd1-4a26-9a2f-9c4e34bd9fcb",
        "terms_url": "https://www.nvidia.com/en-us/about-nvidia/terms-of-service/",
        "width": 700,
        "height": 393,
        "mode": "editorial",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://nvidianews.nvidia.com/news/nvidia-launches-revolutionary-volta-gpu-platform-fueling-next-era-of-ai-and-high-performance-computing",
          "outlet": "NVIDIA Newsroom",
          "published": "2017-05-10",
          "quote": "Equipped with 640 Tensor Cores, V100 delivers 120 teraflops of deep learning performance, equivalent to the performance of 100 CPUs."
        }
      ]
    },
    {
      "id": 1225,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1225",
      "date": "2020-05-14",
      "date_precision": "day",
      "title": "A100 in full production: Ampere unifies training and inference, adopted by every major cloud",
      "summary": "Nvidia said the first Ampere GPU, the A100, was in full production and shipping worldwide, with up to 20 times the performance of its predecessors and adoption by the top cloud providers and server makers. The A100 was the chip the large language models of 2020 to 2022 were trained on, which is why the H100 launched into an installed base that already ran on Nvidia.",
      "type": "launch",
      "entities": [
        "Nvidia"
      ],
      "parties": [
        "nvidia"
      ],
      "published_at": "2026-09-14T03:30:20.279140+00:00",
      "note": "",
      "image": "/images/nodes/nvidia-ai-compute-origins/1225-4acd5114.jpg",
      "image_credit": {
        "credit": "NVIDIA, press image",
        "source_url": "https://nvidianews.nvidia.com/news/nvidias-new-ampere-data-center-gpu-in-full-production",
        "image_url": "https://iprsoftwaremedia.com/219/files/202010/5fb1f5282cfac25847c7e5a4_nvidia-a100-sxm4-80gb/nvidia-a100-sxm4-80gb_85509027-6c18-4236-93e9-16418d375e6f-prv.jpg?v=85509027-6c18-4236-93e9-16418d375e6f",
        "terms_url": "https://www.nvidia.com/en-us/about-nvidia/terms-of-service/",
        "width": 1600,
        "height": 900,
        "mode": "editorial",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://nvidianews.nvidia.com/news/nvidias-new-ampere-data-center-gpu-in-full-production",
          "outlet": "NVIDIA Newsroom",
          "published": "2020-05-14",
          "quote": "NVIDIA today announced that the first GPU based on the NVIDIA® Ampere architecture, the NVIDIA A100, is in full production and shipping to customers worldwide."
        }
      ]
    },
    {
      "id": 1226,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1226",
      "date": "2022-11-30",
      "date_precision": "day",
      "title": "OpenAI launches ChatGPT as a free research preview",
      "summary": "OpenAI released ChatGPT, a conversational model trained with reinforcement learning from human feedback, free to use during its research preview. Its adoption over the following weeks turned generative AI from a research budget into a capital-expenditure race among cloud providers, all of whom were buying the same GPU.",
      "type": "launch",
      "entities": [
        "OpenAI"
      ],
      "parties": [
        "openai"
      ],
      "published_at": "2026-09-14T03:30:20.279186+00:00",
      "note": "",
      "image": "/images/nodes/nvidia-ai-compute-origins/1226-0f363248.jpg",
      "image_credit": {
        "credit": "OpenAI, press image",
        "source_url": "https://openai.com/index/chatgpt/",
        "image_url": "https://images.ctfassets.net/kftzwdyauwt9/7f6c5b76-875f-455d-ebb8813e1022/c16e0d3c81f4b9bcc9d9c65ed69b203d/chatgpt-og.jpg?w=1600&h=900&fit=fill",
        "terms_url": "https://openai.com/brand/",
        "width": 1600,
        "height": 900,
        "mode": "editorial",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://openai.com/index/chatgpt/",
          "outlet": "OpenAI",
          "published": "2022-11-30",
          "quote": "We’ve trained a model called ChatGPT which interacts in a conversational way."
        }
      ]
    },
    {
      "id": 1227,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1227",
      "date": "2023-05-24",
      "date_precision": "day",
      "title": "Nvidia reports \"surging demand\" and says it is significantly increasing supply",
      "summary": "With first-quarter fiscal 2024 data-center revenue at a record $4.28 billion, Jensen Huang said the computer industry was going through two transitions at once, accelerated computing and generative AI, and that Nvidia was significantly increasing supply to meet surging demand. The guidance that followed became the clearest public marker of the H100 shortage.",
      "type": "financial",
      "entities": [
        "Nvidia",
        "Jensen Huang"
      ],
      "parties": [
        "nvidia",
        "jensen-huang"
      ],
      "published_at": "2026-09-14T03:30:20.279232+00:00",
      "note": "",
      "image": "",
      "image_credit": {
        "credit": "",
        "source_url": "",
        "image_url": "",
        "terms_url": "",
        "mode": "licensed",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-first-quarter-fiscal-2024",
          "outlet": "NVIDIA Newsroom",
          "published": "2023-05-24",
          "quote": "“The computer industry is going through two simultaneous transitions — accelerated computing and generative AI,” said Jensen Huang, founder and CEO of NVIDIA."
        }
      ]
    },
    {
      "id": 1228,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1228",
      "date": "2023-08-23",
      "date_precision": "day",
      "title": "Data-center revenue of $10.32 billion, up 171 percent in a year",
      "summary": "Second-quarter fiscal 2024 data-center revenue more than doubled from the prior quarter to a record $10.32 billion, up 141 percent sequentially and 171 percent from a year earlier, as H100 systems shipped to cloud providers and consumer-internet companies. It was the quarter the demand shift first showed in Nvidia's numbers.",
      "type": "financial",
      "entities": [
        "Nvidia"
      ],
      "parties": [
        "nvidia"
      ],
      "published_at": "2026-09-14T03:30:20.279279+00:00",
      "note": "",
      "image": "",
      "image_credit": {
        "credit": "",
        "source_url": "",
        "image_url": "",
        "terms_url": "",
        "mode": "licensed",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-second-quarter-fiscal-2024",
          "outlet": "NVIDIA Newsroom",
          "published": "2023-08-23",
          "quote": "Second-quarter revenue was a record $10.32 billion, up 141% from the previous quarter and up 171% from a year ago."
        }
      ]
    },
    {
      "id": 1229,
      "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1229",
      "date": "2024-02-21",
      "date_precision": "day",
      "title": "Fiscal 2024 closes with data-center revenue of $47.5 billion, up 217 percent",
      "summary": "Nvidia reported fourth-quarter data-center revenue of $18.4 billion, up 409 percent from a year earlier, and full-year data-center revenue of $47.5 billion, up 217 percent. In one fiscal year the data-center business went from a large segment to most of the company, on the back of a single product generation.",
      "type": "financial",
      "entities": [
        "Nvidia"
      ],
      "parties": [
        "nvidia"
      ],
      "published_at": "2026-09-14T03:30:20.279320+00:00",
      "note": "",
      "image": "",
      "image_credit": {
        "credit": "",
        "source_url": "",
        "image_url": "",
        "terms_url": "",
        "mode": "licensed",
        "photographer": "",
        "photographer_url": "",
        "photo_url": "",
        "license": "",
        "license_url": ""
      },
      "sources": [
        {
          "url": "https://nvidianews.nvidia.com/news/nvidia-announces-financial-results-for-fourth-quarter-and-fiscal-2024",
          "outlet": "NVIDIA Newsroom",
          "published": "2024-02-21",
          "quote": "Fourth-quarter revenue was a record $18.4 billion, up 27% from the previous quarter and up 409% from a year ago. Full-year revenue rose 217% to a record $47.5 billion."
        }
      ]
    }
  ],
  "relations": [],
  "links": [
    {
      "from": {
        "event": "nvidia-ai-compute-origins",
        "id": 1225,
        "date": "2020-05-14",
        "title": "A100 in full production: Ampere unifies training and inference, adopted by every major cloud",
        "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1225"
      },
      "to": {
        "event": "nvidia-datacenter-gpu-roadmap",
        "id": 1,
        "date": "2022-03-22",
        "title": "Nvidia announces the Hopper architecture and H100 GPU at GTC 2022",
        "url": "https://whatledto.com/events/nvidia-datacenter-gpu-roadmap#n1"
      },
      "kind": "causes",
      "note": "The A100's widespread cloud and server-maker adoption created the installed base into which Timeline B says the H100 launched."
    },
    {
      "from": {
        "event": "nvidia-datacenter-gpu-roadmap",
        "id": 14,
        "date": "2022-11-16",
        "title": "NVIDIA begins shipping Hopper-based H100 GPUs",
        "url": "https://whatledto.com/events/nvidia-datacenter-gpu-roadmap#n14"
      },
      "to": {
        "event": "nvidia-ai-compute-origins",
        "id": 1228,
        "date": "2023-08-23",
        "title": "Data-center revenue of $10.32 billion, up 171 percent in a year",
        "url": "https://whatledto.com/events/nvidia-ai-compute-origins#n1228"
      },
      "kind": "causes",
      "note": "NVIDIA's start of Hopper-based H100 shipments preceded Nvidia's reported data-center revenue surge, which the later entry says came as H100 systems shipped to cloud providers and consumer-internet companies."
    }
  ],
  "key_node_ids": [
    1222,
    1224,
    1226,
    1229
  ],
  "catalysts": [],
  "open_questions": [
    "Nvidia’s fiscal-2024 disclosures do not break out H100 revenue or unit shipments from revenue generated by other data-center products.",
    "The cited sources show Nvidia increasing supply amid surging demand, but do not quantify the size, duration, or customer allocation of the reported H100 shortage.",
    "The sources document Nvidia’s growth and product milestones but do not provide a direct comparison of its AI-compute market share with competitors."
  ],
  "metrics": [],
  "related": [
    {
      "slug": "nvidia-datacenter-gpu-roadmap",
      "url": "https://whatledto.com/events/nvidia-datacenter-gpu-roadmap",
      "json": "https://whatledto.com/events/nvidia-datacenter-gpu-roadmap.json"
    },
    {
      "slug": "hyperscaler-custom-accelerators",
      "url": "https://whatledto.com/events/hyperscaler-custom-accelerators",
      "json": "https://whatledto.com/events/hyperscaler-custom-accelerators.json"
    },
    {
      "slug": "amd-instinct-ai-gpus",
      "url": "https://whatledto.com/events/amd-instinct-ai-gpus",
      "json": "https://whatledto.com/events/amd-instinct-ai-gpus.json"
    }
  ],
  "changelog": [
    {
      "date": "2026-09-18T03:21:02.436824+00:00",
      "kind": "link",
      "summary": "Linked 1 entry with Nvidia Data Center GPUs: Hopper, Blackwell and Vera Rubin.",
      "node_ids": [
        1225
      ]
    },
    {
      "date": "2026-09-16T08:42:28.460075+00:00",
      "kind": "link",
      "summary": "Linked 1 entry with Nvidia Data Center GPUs: Hopper, Blackwell and Vera Rubin.",
      "node_ids": [
        1228
      ]
    },
    {
      "date": "2026-09-14T03:32:12.747635+00:00",
      "kind": "summary",
      "summary": "Summary, key players and catalysts refreshed.",
      "node_ids": []
    },
    {
      "date": "2026-09-14T03:30:20.301974+00:00",
      "kind": "seed",
      "summary": "Seeded 8 hand-verified node(s) from nvidia-ai-compute-origins.yaml.",
      "node_ids": [
        1222,
        1223,
        1224,
        1225,
        1226,
        1227,
        1228,
        1229
      ]
    }
  ],
  "license": {
    "holder": "What Led To",
    "terms": "https://whatledto.com/methodology#reuse",
    "summary": "Quote entries with attribution and a link to the entry; do not republish whole timelines or train models on the text."
  }
}