Bloomberg Intelligence Estimates 1GW AI Data Center Costs Over $30 Billion

Bloomberg Intelligence reports that constructing a 1GW data center in Abilene, Texas, can cost up to $30.7 billion based on data from Cushman & Wakefield and internal analysis. More than half of this budget, specifically $17.1 billion, is allocated directly to core chips and server hardware. This report highlights the massive capital expenditure required for modern artificial intelligence infrastructure, emphasizing how hardware and power demands are reshaping industry investments. It provides crucial financial insights for stakeholders navigating the surging costs of scaling AI computational power. Beyond the $17.1 billion for chips and servers, the breakdown includes $3 billion for power supply infrastructure, $1.66 billion for cooling systems, and $2.4 billion for construction. Additional costs feature $1.8 billion for land acquisition, $361 million for engineering design, and $2.79 billion for contingency and reserves.