Who would’ve thought that the future of artificial intelligence hinges on... water? That’s exactly the conclusion Google reached when it signed a 20-year, $3 billion agreement with Brookfield to supply up to 3,000 megawatts of hydropower across the U.S.
This isn’t some token green initiative for ESG reporting. It’s the energy backbone powering the company’s AI training and cloud infrastructure where data centers churn nonstop, teaching machines to think and ensuring digital services stay reliably online.
Officially titled the Hydro Framework Agreement, the deal is anything but boring. Over the next two decades, Google will draw electricity from two revitalized hydroelectric plants in Pennsylvania Holtwood and Safe Harbor which together generate 670 megawatts. But the potential goes far beyond that. The contract allows expansion up to 3,000 megawatts from other modernized hydropower sites.
Why hydro? Because unlike solar panels that sleep at night or wind turbines that might call in sick, water keeps flowing consistently and reliably. And in the world of AI, where millions of operations occur every second and servers heat up faster than TikTok trends expire, stability is everything.
Also Read: Google Lets OpenAI Use Its Cloud – A Bold Move Between AI Rivals in 2025
This move isn’t a one-off. In 2024 alone, Google signed power purchase agreements (PPAs) for 8 gigawatts of renewable energy. Data center emissions have already dropped by 12%, but clearly, that’s not enough for Google. The next bold step? Starting in 2035, the company will become the first buyer of nuclear power under a dedicated energy contract.
While others debate whether wind or solar is greener, Google is buying it all going all-in on every viable renewable.
The goal is ambitious yet refreshingly specific: reach net-zero emissions by 2030. Not “we aim to,” not “we’ll explore” but a real, measurable commitment. Not just in words, but in megawatts, gigawatts, and the language of actual engineering.
