Enterprises stuck with 5% GPU utilization due to FOMO — impacts cloud compute pricing

Context

Enterprises are running their GPU fleets at 5% utilization due to fear of losing allocation, leading to wasted resources and higher cloud compute pricing. This dynamic is driving a split in the cloud market, with commodity providers offering lower prices for workloads that can live there. Relevant for companies with significant AI workloads evaluating cloud compute providers.