TRACING THE MONEY: Private Equity, Private Credit, and the Data Center Bubble - Americans for Financial Reform

An Americans for Financial Reform report titled "Tracing the Money: Private Equity, Private Credit, and the Data Center Bubble," authored by Iván Cazarin and Dustin Duong and published in October 2026, examines the role of private equity in financing the U.S. data center buildout. The report estimates that as of December 2025, 78 private equity firms backed roughly 13 to 17 gigawatts of active data center capacity across 670 data centers or campuses, equal to 25 to 30 percent of the 54 GW of total active U.S. capacity. Private equity-owned or -financed companies have also announced or initiated at least another 26.8 GW in development, under construction, or actively planned, which the report says amounts to over half of the 50 GW pipeline projected by 2030.

The concentration is pronounced in certain states. According to the report, private equity firms own one third of active data center capacity in Texas, nearly half of current capacity and one third of planned expansion in Arizona, and nearly one quarter of online capacity and over 10 percent of projects in development in Virginia. The five largest private equity data center owners by total active and planned capacity are Brookfield Asset Management, Ares Management, Blue Owl Capital, DigitalBridge, and Pantheon Ventures, which together hold investments in 17 percent of existing capacity and 30 percent of planned capacity. By the end of 2025, the report notes, private equity firms backed more than 200 data center companies worldwide, and Blackstone said its portfolio company QTS was the largest data center operator in the world as of June 2025.

The report draws on a Morgan Stanley estimate that of the $2.9 trillion expected to be spent on new data centers between 2025 and 2028, about $1.1 trillion, more than a third, will come from private equity through private credit loans, joint ventures, or equity stakes, while Big Tech firms are shifting from owning capacity to leasing it. The authors argue this structure moves expansion costs off Big Tech balance sheets and obscures leverage. The report also cites a projection that one third of the roughly $3 trillion in data center spending through 2028 will be debt-financed, and it warns that the sector could pose financial stability risks if AI revenue and usage fall short of projections. It cites a March 2025 McKinsey finding that more than 80 percent of businesses adopting generative AI report no profitability impact, and it raises concern that public pension and university endowment capital is exposed to the sector.

The report also highlights community and fiscal costs, including land consumption, water use, utility cost increases, and emissions. It cites Georgia's loss of $2.5 billion in tax receipts in a single year due to data center tax incentives. The authors' framing is advocacy-oriented, and their conclusions on bubble risk and financial exposure are their own analysis rather than established fact.