A highly specialized compute environment designed to run artificial intelligence training and inference workloads. Because of their high cost, AI clusters require continuous, high utilization and rapid reprovisioning of tenants to maximize return on investment.