SkyPilot, founded by Databricks cofounder Ion Stoica, raised $20 million to make “AI compute” switching easier across hardware and cloud providers. The company positions itself as a neutral orchestration layer that helps labs and enterprises run workloads across different GPU supply sources without engineering delays. The startup story ties to a broader enterprise pain point: today’s labs can call multiple cloud providers early to assemble enough GPUs, but aggregating those resources and using them efficiently remains costly and operationally hard. For higher education research centers and institutional AI labs, orchestration platforms can affect both budgeting and performance. If compute utilization improves, the same GPU spend can produce more experiments, faster iteration cycles, and potentially lower barrier costs for distributed projects. The funding also highlights investor attention on the “next layer” of the AI stack—moving beyond model access into infrastructure management and utilization efficiency.