AI system computational resources

Description

Adequate computing infrastructure for AI training, testing, and operation with capacity planning and environmental impact consideration.

Implementation Guidance

Infrastructure Assessment

Inventory current compute resources (GPU/TPU clusters, cloud allocations, storage) and map to AI workload requirements. Identify capacity gaps for training, inference, and testing workloads. Include disaster recovery and failover planning for production AI systems.

Environmental Impact

Track energy consumption and carbon footprint of AI training and inference operations. Set efficiency targets and document optimization strategies (model distillation, efficient architectures, batch scheduling). Report environmental metrics to stakeholders.

Capacity Planning

Establish forward-looking capacity plans aligned with the AI development roadmap. Include cost projections, scaling triggers, and resource allocation governance to prevent uncontrolled compute spend.

Evidence Requirements

• Infrastructure inventory and capacity assessment
• Resource allocation plans and budgets
• Energy consumption and carbon footprint reports
• Capacity planning documentation with forecasts