Microsoft Corp.’s purchase of nearly 400 megawatts of solar power from EDP Renewables North America this week speaks not only of its effort to reduce carbon emissions in a few years but the soaring energy demands it must meet now to achieve its artificial intelligence (AI) and cloud dreams.
The additional power, which adds about 2% to Microsoft’s nearly 20 gigawatts of renewable energy capacity, covers three solar projects developed by EDP in Illinois and Texas. Microsoft now has a total of five projects with EDP in the U.S.
“It’s exciting when projects achieve commercial operations and begin generating clean energy in support of grid decarbonization,” Kourtney Nelson, senior director of renewable energy procurement at Microsoft, said in a statement. Microsoft has vowed to become carbon negative by 2030.
To that end, Microsoft last month said it had bought more than 7 million tons of carbon credits from Chestnut Carbon – or enough to cover half its emissions in 2023.
The software giant, like its Big Tech brethren, has poured tens of billions of dollars into AI data center development over the past year, while seeking cleaner energy sources. Its expanding portfolio includes not just solar but nuclear energy facilities. The company has agreed to pay Constellation Energy to resuscitate the shuttered Three Mile Island nuclear power plant in Pennsylvania, and said it will buy power from Helion Energy, a Seattle startup that seeks to build the world’s first nuclear fusion power plant by 2028.
As AI dominates economic and political policy across North America, Europe, and Asia, so has its energy use and demand for more data centers. Management consultant McKinsey says the need for data centers worldwide could grow by 240GW (some 240 billion gigawatts) between 2023 and 2030, with the U.S. accounting for 12% of all power generated.
DeepSeek’s recent debut as an affordable, high-performance AI model is likely to be a catalyst in increased AI use, spurring even more energy demands from enterprises and consumers in the U.S. and abroad. The Trump administration said a $500 billion AI infrastructure initiative with OpenAI, Oracle Corp., and SoftBank will be dedicated to building large data centers in the U.S. over the next four years.
Traditionally non-tech companies like Chevron Corp. and Exxon have joined the power grab to produce energy sources for AI. Chevron’s recent joint development project with Engine No. 1 and GE Vernova initially plans to deliver four gigawatts – enough to power 3 million to 3.5 million U.S. homes – via co-located data centers in the Southeast, Midwest, and West by late 2027.
