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

The data center economy functions as the physical infrastructure of intelligence, enabling the proliferation of AI by providing the necessary power, cooling, and compute capacity. This massive capital investment—often reaching $60 million per megawatt—represents a fundamental shift in the global economy by creating "digital labor," which can be scaled far more rapidly than traditional human workforces. Crusoe CEO Chase Lochmiller emphasizes an energy-first strategy, co-locating high-performance computing clusters with abundant, low-cost energy sources like wind and natural gas in regions like West Texas. While compute hardware remains a critical component, the primary bottleneck gating AI growth is the availability of energized, ready-to-use data center shells. This infrastructure-heavy model aims for a four-year payback period, with potential for further margin expansion through managed AI services that abstract away the underlying hardware complexities.