Bloomberg Intelligence: 1GW AI data center could cost over $30B
Bloomberg Intelligence published a briefing on AI data center costs that, drawing on Cushman & Wakefield data and its own analysis, estimates building a 1GW facility in Abilene, Texas could cost about $30.7 billion (roughly RMB 205.87 billion). Chips and server hardware account for $17.1 billion, more than half the total, followed by power supply infrastructure at $3.0 billion, cooling at $1.66 billion and networking and cabling at $1.07 billion. The remainder includes $2.4 billion construction, $2.79 billion contingency and reserves, $361 million design and engineering, $382 million miscellaneous, $180 million land and $175 million commissioning.
The itemized breakdown gives a rare view of where AI data center capital goes, showing that hardware rather than the building itself dominates the cost of expanding compute capacity.