Announcing Anyscale on Azure public preview: Powered by Ray on AKS (opens in new tab)

What's new The defining workload of the cloud era is no longer a web service or a database. It is a distributed AI training or inference job that needs thousands of GPUs, a coherent data plane, and a way for a handful of engineers to operate all of it without losing their weekends. Getting there is still harder than it should be. GPU capacity is fragmented across regions and SKUs, data is scattered across the lifecycle, and identity and credentials sprawl with every cluster a team adds. Today...