The rapid growth of AI workloads presents both an opportunity and challenge for public cloud providers. While the overall cloud market is projected to reach $2 trillion by 2030, with AI driving significant growth, cost considerations may limit cloud providers' ability to capture this opportunity. Infrastructure costs for AI workloads, particularly those requiring specialized GPU resources, are substantially higher in public clouds compared to traditional data centers and colocation facilities. This cost differential is especially pronounced for inference workloads, which could represent up to 90% of AI compute by 2030 and can be up to 75% cheaper when run on-premises.