AI's growing need for computing power is expected to significantly expand the scale of data centers over the next five to ten years. Currently, large data centers operated by hyperscale cloud service providers range from 50 megawatts to over 200 megawatts. However, the demands of AI may push these facilities to explore capacities of 1 gigawatt and higher, which will have substantial implications for infrastructure, power production, and cooling systems.

AI is set to transform edge computing by necessitating the use of smaller, domain-specific language models that can handle tasks requiring lower latency. This shift will lead to innovations in user device form factors and computing capabilities, allowing for more efficient processing closer to the edge, particularly in environments with low tolerance for latency, such as autonomous driving.

The expansion of large data centers to accommodate AI's needs will present several challenges, including increased power consumption and competition for resources. As utilities face rising requests for electrical capacity, the demand will compete with other sectors, such as electric vehicles. Additionally, the construction of these facilities requires significant labor, which may strain the workforce, particularly in specialized areas like electrical and cooling systems.