The option to reserve instances and GPUs for inference endpoints may help enterprises address scaling bottlenecks for AI workloads, analysts say. AWS has launched Flexible Training Plans (FTPs) for ...