{
    "action": "final",
    "content": "## Dataset Overview\n\nThis dataset is a **comprehensive global registry of AI supercomputing infrastructure**. It catalogs 482 specialized computing systems deployed worldwide specifically for artificial intelligence workloads—the massive machines that power large-scale AI research, model training, and development.\n\n## What a Single Row Represents\n\nEach row represents **one distinct AI supercomputer installation**. This is not transactional data tracking usage or events; rather, it's a dimensional snapshot capturing the specifications, capabilities, and context of individual machines at a point in time.\n\nFor each system, the dataset records:\n\n- **Technical specifications**: Computing power (measured in operations per second at different precisions: 8-bit, 16-bit, 32-bit), chip types and quantities, energy efficiency\n- **Infrastructure details**: Power capacity, hardware costs, H100 equivalents (a standardization metric)\n- **Organizational context**: Owner, users, sector, country, and precise geographic coordinates\n- **Temporal information**: When the system became operational, when it was decommissioned (if applicable)\n- **Competitive positioning**: Rank when first operational, percentage of the largest cluster at that time, whether it built upon or was superseded by other systems\n\n## Business Purpose\n\nThis asset provides visibility into the **competitive landscape of AI computational capability**. It answers questions like: Which organizations possess the hardware necessary to train frontier AI models? How has supercomputing capacity evolved over time? Where is AI infrastructure concentrated geographically? What does it cost to compete at the cutting edge?\n\nThe dataset operates as a standalone intelligence resource—there are no joins to other tables, reflecting its role as a self-contained reference catalog rather than part of an operational system."
}