Microsoft and Brookfield sign $10 billion green energy deal to meet artificial intelligence power needs


In 2020, Microsoft made a compelling commitment to become carbon negative by 2030. Remove all carbon emissions from the environment”. Since its founding in 1975, the company has emitted carbon directly or through electricity consumption. “

Then there’s artificial intelligence.

Artificial intelligence is already consuming vast amounts of electricity, changing the equation for tech companies balancing the twin priorities of managing carbon emissions and remaining competitive in the AI ​​race. The data centers that store information and host large-scale language models trained for artificial intelligence are equipped with high-performance computer chips that require large amounts of energy and large amounts of water to cool.

Some analysts predict that AI could consume a quarter of all U.S. electricity by 2030. The current power grid cannot handle such a large surge in demand, so tech companies are taking matters into their own hands.

This morning, Microsoft announced a $10 billion agreement with asset management company Brookfield to provide Microsoft with up to 10.5 gigawatts of renewable energy generation capacity between 2026 and 2030. agreement, roughly equivalent to the electricity required to generate electricity.

“It’s a huge difference,” said Xavier Smith, director of research, energy and industry at research firm AlphaSense. wealth. “Energy companies… used to be excited about four-megawatt deals. Then they turned into 600-megawatt deals. Now we’re in the gigawatt range. It’s growing at an exponential rate.

Brookfield is a giant in renewable energy. It manages more than 7,000 power generation facilities on five continents, ranging from wind farms to solar to hydropower. But even Brookfield may need to build more capabilities to meet Microsoft’s needs.

That could come in the form of more solar projects, which could be a good fit given the potential Sunbelt solar sites’ proximity to growing data centers in Phoenix and Las Vegas. Nuclear power has emerged as another possible option for the tech industry: Amazon last month purchased a nuclear-powered data center site in Pennsylvania, and OpenAI CEO Sam Altman backed a company seeking to use small nuclear reactors to power Startups powered by data centers.

“Nuclear energy is increasingly becoming an important part of this conversation,” Smith said. “As early as last fall, [Microsoft] There is a job posting for a Nuclear Technology Manager. They wanted someone to develop a strategy for implementing small modular reactors for them. As a result, people are increasingly thinking about nuclear energy for the future of this field.

Neither company specified what types of renewable energy Brookfield would provide Microsoft, nor the specific financial terms of the agreement.this Financial Times Based on current industry trends and pricing structures, the deal is expected to be worth approximately $10.5 billion.

Renewable energy prices have fallen significantly in recent years and are likely to continue falling, but Smith said it still makes sense for companies like Microsoft to lock in contracts now rather than wait for the possibility of cheaper power in the future or try to establish themselves energy company. Because artificial intelligence is scaling so quickly and energy projects can take years to build from proposal to completion, tech companies have an incentive to move quickly.

“I actually think they would be wise to lock this in right now,” Smith said. “All the hyperscale enterprises are saying that by 2030, their data centers will be running on 100% green energy… The fastest way to achieve that is to work with the people who have already built this infrastructure, rather than trying to outsource the field. delivery time, starting from scratch and constantly developing yourself.

Microsoft manages more than 300 data centers around the world, which are used to store information and train large language models to support chatbots such as ChatGPT and artificial intelligence applications such as Copilot. Due to the decentralized nature of AI infrastructure, some have suggested that off-grid customized energy projects that directly power data centers may be a more efficient way to power AI. While it might be cheaper for Microsoft or its competitors to build their own off-grid solution, Smith said that’s a less attractive option.

“although [tech companies] They want access to renewable energy, they don’t want to run power centers,” Smith said. “Google doesn’t want to run a power [plant]. Microsoft doesn’t want to operate nuclear facilities.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top