Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Building AI Factories

The data center economy serves as the physical infrastructure backbone for the current AI boom, transforming energy into digital labor to accelerate GDP growth. As hyperscalers commit hundreds of billions to infrastructure, the primary bottleneck has shifted from chip availability to the deployment of energized, high-density data centers. Chase Lochmiller, CEO of Crusoe, emphasizes an energy-first strategy, co-locating computing clusters with abundant, low-cost power sources like wind and natural gas to bypass traditional grid constraints. While building these facilities requires massive capital expenditure—roughly $60 million per megawatt—the integration of managed services and inference scaling creates a viable path to profitability. Ultimately, the industry is shifting toward modular, vertically integrated designs to overcome labor shortages and infrastructure complexities, aiming to sustain the rapid expansion of AI compute capacity while navigating evolving economic and technical demands.