AWS has already made improvements. In the past, if you provisioned a large EC2 instance and underutilized it, you still paid a premium. Now, AWS encourages using Kubernetes, where you primarily pay for the compute power you actually use in production. There is room for improvement. You might end up paying a high price if you're not careful and you provision a server that's underutilized. AWS has left it to engineers to figure out solutions. If you find the cost too high, you can move to Kubernetes, which might be a better solution for you than large EC2 instances. So, the improvements need to come from the user side, not the provider. Software engineers and engineering teams need to know their limits with EC2 instances. They need to recognize when it's time to transition their applications to Kubernetes. This means building with the cloud in mind from the start, making it easier to move solutions to the cloud without suffering upgrades and integration issues.