busi488energy/data/ai-milestones.json

[
  {
    "date": "2020-06-11",
    "title": "GPT-3 API Launch",
    "description": "OpenAI opens GPT-3 API access, demonstrating that scaling language models yields emergent capabilities. Signals the start of the large-model era.",
    "category": "model_launch"
  },
  {
    "date": "2021-06-29",
    "title": "GitHub Copilot Preview",
    "description": "GitHub launches Copilot technical preview, the first mass-market AI coding assistant. Powered by OpenAI Codex, it demonstrates commercial viability of large language models.",
    "category": "model_launch"
  },
  {
    "date": "2022-04-06",
    "title": "DALL-E 2 Launch",
    "description": "OpenAI unveils DALL-E 2, a breakthrough in AI image generation. Kicks off the generative AI media wave and accelerates GPU demand for inference workloads.",
    "category": "model_launch"
  },
  {
    "date": "2022-08-22",
    "title": "Stable Diffusion Open-Source Release",
    "description": "Stability AI releases Stable Diffusion publicly under an open-source license, democratizing AI image generation and driving massive consumer GPU demand.",
    "category": "model_launch"
  },
  {
    "date": "2022-11-30",
    "title": "ChatGPT Launch",
    "description": "OpenAI releases ChatGPT, reaching 1 million users in 5 days and 100 million in 2 months. Sparks the generative AI boom and triggers an unprecedented race for GPU compute.",
    "category": "model_launch"
  },
  {
    "date": "2023-01-23",
    "title": "Microsoft Invests $10B in OpenAI",
    "description": "Microsoft announces a $10 billion investment in OpenAI, the largest single AI investment at the time. Signals hyperscaler conviction that AI will drive massive infrastructure spend.",
    "category": "market"
  },
  {
    "date": "2023-03-14",
    "title": "GPT-4 Launch",
    "description": "OpenAI releases GPT-4, a multimodal model accepting text and image inputs. Demonstrates a step-change in AI capability, intensifying the arms race for training compute.",
    "category": "model_launch"
  },
  {
    "date": "2023-05-30",
    "title": "NVIDIA Hits $1 Trillion Market Cap",
    "description": "NVIDIA becomes the first chipmaker to reach a $1 trillion market capitalization, driven by explosive AI GPU demand. Stock price reflects the market's bet on sustained datacenter buildout.",
    "category": "market"
  },
  {
    "date": "2023-07-18",
    "title": "Meta Releases Llama 2 Open-Source",
    "description": "Meta releases Llama 2 for research and commercial use, open-sourcing 7B, 13B, and 70B parameter models. Accelerates AI adoption by lowering barriers and increasing total compute demand.",
    "category": "model_launch"
  },
  {
    "date": "2023-08-17",
    "title": "NVIDIA H100 GPU Shortage Peak",
    "description": "Reports confirm H100 GPUs are sold out through Q1 2024 with 6+ month lead times. CoWoS packaging bottleneck limits supply to ~550K units for the year, driving datacenter expansion plans.",
    "category": "infrastructure"
  },
  {
    "date": "2023-10-30",
    "title": "Biden Executive Order on AI Safety",
    "description": "President Biden signs Executive Order 14110, the most comprehensive US AI governance action. Establishes safety standards, reporting requirements for large model training runs, and energy impact assessments.",
    "category": "policy"
  },
  {
    "date": "2023-12-06",
    "title": "Google Gemini Launch",
    "description": "Google announces Gemini 1.0 with Ultra, Pro, and Nano variants. Google's entry into the frontier model race further escalates industry-wide datacenter investment.",
    "category": "model_launch"
  },
  {
    "date": "2024-03-04",
    "title": "Claude 3 Family Launch",
    "description": "Anthropic releases Claude 3 Haiku, Sonnet, and Opus, the first AI model family with clear speed/cost/capability tradeoffs. Broadens enterprise AI adoption and inference compute demand.",
    "category": "model_launch"
  },
  {
    "date": "2024-03-18",
    "title": "NVIDIA Blackwell GPU Architecture Announced",
    "description": "NVIDIA unveils Blackwell B200 and GB200 GPUs at GTC 2024. Each GB200 rack draws up to 120 kW, signaling that next-generation AI hardware will dramatically increase datacenter power density.",
    "category": "infrastructure"
  },
  {
    "date": "2024-05-13",
    "title": "GPT-4o Launch",
    "description": "OpenAI releases GPT-4o (omni), a natively multimodal model processing text, image, and audio. Made free in ChatGPT, massively expanding inference volume and datacenter load.",
    "category": "model_launch"
  },
  {
    "date": "2024-07-15",
    "title": "PJM Capacity Auction Sets Record Prices",
    "description": "PJM Interconnection's capacity auction clears at $16.1 billion, up 9.5% year-over-year. Data center load growth identified as the primary driver, with 30 GW of new demand projected by 2030.",
    "category": "market"
  },
  {
    "date": "2024-12-18",
    "title": "NERC Warns of Nationwide Capacity Shortfalls",
    "description": "NERC's Long-Term Reliability Assessment warns that all US regions except MRO-Manitoba face elevated or high risk of supply shortfalls by 2028, driven primarily by datacenter load growth.",
    "category": "policy"
  },
  {
    "date": "2025-01-03",
    "title": "Microsoft Commits $80B to AI Datacenters",
    "description": "Microsoft announces plans to spend $80 billion on AI-enabled datacenters in fiscal year 2025, more than half in the US. Largest single-company infrastructure commitment in history.",
    "category": "infrastructure"
  },
  {
    "date": "2025-01-21",
    "title": "Stargate Project Announced ($500B)",
    "description": "OpenAI, SoftBank, and Oracle announce the Stargate Project: $500 billion over four years for AI infrastructure, with $100 billion deployed immediately. Represents a new scale of AI investment.",
    "category": "infrastructure"
  },
  {
    "date": "2025-11-12",
    "title": "Anthropic Announces $50B Infrastructure Buildout",
    "description": "Anthropic announces plans to spend $50 billion on US AI infrastructure, starting with custom datacenters in Texas and New York. Signals that AI labs themselves are becoming major power consumers.",
    "category": "infrastructure"
  }
]