Primary Computing Use of Data Centers in Santa Clara
Estimated Based on Total Data Center Electrical Load (MW) - 2026 Al Training 16%
• Teaching an Al model by processing large amounts of data so it can learn patterns and develop its capabilities.
• Example: Training a model to understand language or recognize images. 16%
Al Inference 13%
• An evolution of Al development, where a trained artificial intelligence model processes new, unseen data to generate real-world predictions, decisions, or 13% outputs.
• Example: When an autonomous Al agent executes multi-step workflows, makes decisions, and uses tools independently to achieve a specific goal without constant human intervention.
71% All Other Data Centers 71%
• Cloud infrastructure (laaS/PaaS)
• Enterprise IT / Saas
Semiconductor / HPC / R&D
Network / CDN / Telecom
Disaster recovery / storage / other
Note: Planning-level estimate based on publicly identifiable facilities, operator disclosures, market information, and workload characteristics.
Percentages represent the primary computing workload of stand-alone data centers. Actual workloads may overlap and change over time.