Executive Summary Microsoft’s recent capital expenditure adjustments underscore a pivotal shift in the AI market, as the primary focus transitions from model training to inference. Distributed inference is emerging as a significant yet underappreciated demand driver. The company’s decision to delay certain data center construction projects signals a strategic recalibration in response to evolving market structures, a trend mirrored by Google, Amazon, and Meta. However, Microsoft’s fungible data center concept stands out as a key