Provision a fixed number of AI resources regardless of workload demand and do not implement any scaling mechanisms.

Manually scale the AI service up or down based on workload demand, but do not implement any automation or auto-scaling mechanisms.

Implement auto-scaling based on CPU utilization and configure scaling policies to scale the AI service up or down based on demand.

Implement auto-scaling based on a fixed schedule regardless of actual workload demand, and ignore any performance metrics or cost implications.