The world of cloud computing is currently dominated by Amazon Web Services (AWS) which has left Google's Cloud Platform and Microsoft's Azure service playing catch up. In an effort to stand out from the competition, Azure has partnered with Cray to add their supercomputers to the Azure cloud service. The goal is to attract users with high performance computing (HPC) workloads that would have otherwise gone with dedicated hardware.
While some workloads may require higher end systems with FPGAs and GPUs, many cloud computing applications can be accomplished on relatively low-tier virtualized computers. Moving beyond that though, there is a growing demand for true supercomputer performance in the cloud. These include use cases like climate science, medicine, artificial intelligence, and scientific research. This level of computing power doesn't come cheap and thus many users haven't been able to make use of it until now.
Cray will be bringing in their XC and CS supercomputers as well as the ClusterStor storage system to the Azure datacenters. They will be directly connected on the Azure network which will allow for easy integration with Azure virtual machines, data storage, machine learning systems, and artificial intelligence platform.
Unlike traditional cloud computing services, these systems will only be available for provisioning as dedicated resources. That means you won't be able to timeshare or dynamically scale your performance as easily. The advantage of going to the cloud is that it lets users access top of the line hardware without the cost and maintenance associated with housing it on premises.