North America's leading cloud providers, including Microsoft, Amazon, and Google, are transitioning to a Pre-Provisioned Throughput Unit (PTU) model to address soaring AI token costs. This shift comes as these companies report significant revenue growth but face stock declines due to escalating AI expenses. Microsoft's Intelligent Cloud division reported over $50 billion in quarterly revenue, while Amazon's AWS and Google Cloud also posted strong growth. However, their stocks fell due to concerns over unsustainable AI costs. The PTU model allows customers to pre-purchase computing power at a fixed rate, offering cost predictability compared to the traditional token-based billing. This change aims to stabilize revenue growth by encouraging increased AI usage without fear of cost overruns. Microsoft, AWS, and Google are adopting distinct strategies to implement PTU, focusing on ecosystem bundling, cost advantages, and performance premiums, respectively. This transition is expected to impact upstream chip manufacturers and downstream AI application providers, reshaping the AI industry's economic landscape.