With Amazon's Web services, you've been able to store stuff in S3 or SimpleDB. You've also been able to fire up as many machine instances as you liked with storage that went away when you shut the machine down. Anything you wanted saved better be in a database somewhere else, or you had to painstakingly copy it out to S3 yourself. Last night Amazon announced persistent storage on EC2. Now you can create disks in S3 and attach them to EC2 instances. You want a terabyte of storage for your machine, just create it in S3 and mount it.
Another feature rolled out last night is snapshots. Need backup or the ability to rollback? Snapshot the instance and it's on S3, ready to use. You can create new volumes from any particular snapshot.
These two features make Amazon's grid computing platform a very nice place for startups to experiment, develop, and build out. All with little or no capital cost.