YellowDog Insights

YellowDog Insights gives engineering and infrastructure teams up-to-date visibility of global cloud compute: more than 10,000 GPU and CPU instance type and region combinations across AWS, benchmarked, ranked, and searchable in one place, to help you plan where to run workloads using your existing cloud accounts. GPU scarcity is a location problem. When preferred silicon runs short in one region, most teams only find out by checking their own cloud account, one region at a time. Insights removes that blind spot by showing every region where the instance type you want is offered. It does not indicate live cloud capacity; provisioning availability is determined by the YellowDog platform at runtime. Every instance type is benchmarked for CPU and memory performance using QuantLib and sysbench, and ranked by price, benchmark score, and price/performance, alongside sustainability data. A broad set of filters lets you zoom in on exactly what you care about: processor architecture, GPU model, memory, operating system, region, and spot. Use Insights to discover where the instance types you need for GenAI training and inference are offered, identify the most cost-effective options for batch and HPC workloads, compare spot across regions, or confirm which regions list a specific GPU family before committing to an architecture. Insights integrates with the YellowDog platform: once you've identified where your preferred hardware is offered, YellowDog checks live cloud capacity and orchestrates provisioning across your existing AWS accounts.