Carlyle is investing heavily in AI infrastructure by developing large-scale energy campuses, like those built by Copia Power, that integrate diverse power sources—including solar, storage, and natural gas—directly adjacent to data centers to ensure reliable, cost-effective energy supply. This comprehensive, long-term approach addresses critical bottlenecks such as power access and aims to build multiple massive campuses in the western U.S. to support the growing demands of AI computing.
The discussion centers on the significant investment opportunity presented by AI infrastructure from the perspective of infrastructure investors like Carlyle. They emphasize a piece-by-piece approach with a long-term horizon, focusing on developing AI infrastructure assets comprehensively across the entire value chain. This approach includes not only building data centers but also addressing critical bottlenecks such as access to power, which is essential for the efficient operation of these centers.
A key point highlighted is the challenge of energy supply for data centers, with reference to Elon Musk’s comments on energy being a major bottleneck for AI development. Carlyle is tackling this by creating large-scale energy campuses that integrate power generation capacity directly adjacent to data centers. This strategy ensures timely, cost-effective, and reliable power delivery, which is crucial for the high demands of AI computing infrastructure.
One notable initiative discussed is Copia Power, a company formed by Carlyle in 2021 to build these expansive energy campuses. These campuses function like mini-cities, combining various energy sources including solar, storage, and natural gas. The integrated approach connects these campuses to the grid, avoiding isolated systems and enhancing overall reliability and efficiency.
The conversation also touches on the realistic energy mix needed to support AI infrastructure. While renewable energy sources like solar and storage are vital, natural gas remains an important component to ensure consistent and reliable power. This balanced “all of the above” energy strategy is necessary to meet the timing, cost, and reliability requirements of large-scale AI data centers.
Finally, the scale of these projects is immense, with one Copia site in Arizona being twice the size of Manhattan and involving around $20 billion in capital investment for both power generation and data center assets. Carlyle plans to develop multiple such campuses in the western United States, strategically located near demand centers to optimize cost, delivery time, and reliability, thereby supporting the growing needs of AI infrastructure development.