With the rapid growth of AI technology and cloud computing demand, Google announced that it would invest up to 30 billion pounds (about NT$1200 billion) to cooperate with Brookfield Asset Management's renewable energy division.Reaching an Agreement, will obtain up to 3000 megawatts (MW) of hydropower supply, becoming the largest hydropower procurement case in history.
The Hydro Framework Agreement (HFA) collaboration will initially see Brookfield's Holtwood and Safe Harbor power plants in Pennsylvania, USA, provide 670 megawatts of carbon-neutral electricity to power Google's growing AI and cloud data centers. Google will also have rights to upgrade and develop these facilities, expanding grid energy supply over the long term.
Currently, Google is focusing on the PJM Interconnection, the largest power grid in the United States. This grid, which covers more than 6500 million users, is facing challenges in meeting the power supply demands of the surging computing power of large technology companies. In the future, the two parties may expand the scope of their cooperation to other regions in the United States.
"We are committed to responsibly expanding the digital infrastructure that supports people, communities, and businesses every day," Google said in a statement.
This energy procurement project coincides with Meta's recentThe announcement of investing hundreds of billions of dollars in AI data centers overlaps with the timingMeta plans to build multiple hyperscale data centers. The first, called Prometheus, is expected to open in Ohio next year. It will be followed by the Hyperion data center, which is said to be "almost as large as Manhattan Island." The center will be expanded to 5 gigawatts (GW) of capacity within a few years, which is expected to set a new global data center scale record.
However, the expansion of large-scale AI data centers also brings another focus: water consumption.
According to the New York Times, a typical data center consumes approximately 50 gallons of water per day, while next-generation AI centers can consume millions of gallons daily. For example, the Meta Data Center, located east of Atlanta, Georgia, has caused local well water levels to drop, municipal water prices to skyrocket, and is expected to face water rationing within the next five years.
"Data centers are draining the wealth of the entire community," said Mike Hopkins, water director for Newton County, Georgia. "We simply don't have enough water." Similar problems are emerging in data center hotspots such as Texas, Arizona, Louisiana and Colorado. Some areas have even suspended residential development due to water shortages.
At a time when AI is driving infrastructure reconstruction, technology giants are facing not only a race for energy computing, but also the long-term challenge of how to obtain and manage limited natural resources.
