The scale of artificial intelligence infrastructure construction in the United States has reached unprecedented levels, with both the number of hyperscale data centers and the planned capital investments surging as technology firms race to expand capacity for growing AI demand.
According to an analysis from independent macroeconomic research firm The Kobeissi Letter, the US currently operates approximately 90 data centers with capacities of 100 megawatts or more, a record high. If all planned projects come to fruition, that number could more than triple to roughly 280 facilities by 2030.
This expansion is accompanied by a massive investment cycle. Tech companies are projected to pour around $7 trillion into data center infrastructure by 2030, underscoring the enormous scale of the facilities needed to support AI workloads. The hyperscalers have already committed significant capital, with Amazon, Microsoft, Alphabet, and Meta having signed data center lease agreements totaling more than $1.5 trillion since the AI boom began, including contracts not yet in effect.
This construction wave extends well beyond the hyperscalers, creating broad opportunities across the entire AI infrastructure supply chain. Semiconductor and networking companies are ramping up investments to meet escalating compute demands, while server, power, and data center operations firms are simultaneously expanding their capacity.
Given the scale of planned capacity and investment, Kobeissi noted that the AI buildout represents one of the largest capital expenditure cycles in history. Globally, the scale of AI data center spending is equally striking. A recent report from PwC projects cumulative worldwide data center expenditures will reach $31.6 trillion by 2050, with the US accounting for nearly half of that figure at approximately $15.1 trillion, followed by the Asia-Pacific region at $8.2 trillion.
Should AI adoption accelerate faster than expected, data center spending over the next 25 years could even reach $50 trillion, potentially dwarfing previous capital expenditure cycles tied to railroads, the internet, and electrification.
Comments