Microsoft AI Wave Challenging Its Climate Commitment

Microsoft's climate pollution is now much higher than its bold climate commitments in 2020. In fact, in the 2023 fiscal year, its greenhouse gas emissions were about 30% higher, indicating how challenging it is for the company to achieve its climate goals while also striving to become a leader in the AI field.

Training and running AI models is an increasingly energy-intensive activity, and its impact on the climate is just beginning to emerge. Microsoft's latest sustainability report is a typical case of the challenges faced by large tech companies that have made a series of climate commitments in recent years, which may increase pollution due to their focus on AI.

As early as 2020, Microsoft set a goal to achieve carbon neutrality by the end of this century. In other words, it pledged to reduce greenhouse gas emissions by more than half and then capture more carbon dioxide emissions than it produces. Considering that carbon capture technology was just beginning to emerge at that time, it was a bold commitment. The company also needs to promote the deployment of more renewable energy in its operations' power grid.

Now, Microsoft's recent emphasis on AI seems to make achieving this goal even more challenging. So far, Microsoft has invested over $13 billion in OpenAI and, with the help of the new feature Copilot in Microsoft 365, aims to "turn everyone into a prompt engineer for generative AI."

"In 2020, we announced the so-called 'Carbon Moonshot.' That was before the explosive growth of artificial intelligence," said Microsoft President Brad Smith in an interview with Bloomberg. "So, in many ways, if you just look at our own predictions about the expansion of AI and its power needs, the distance to achieving this goal is five times further than it was in 2020."

A closer look at the data in Microsoft's sustainability report reveals how far the company has gone in the wrong direction. In the past fiscal year, it emitted 15.357 million tons of carbon dioxide, equivalent to the annual carbon pollution of Haiti or Brunei.

Data centers used for training AI are more energy-intensive than traditional data centers, which already consume a significant amount of electricity to run servers and cooling systems to prevent overheating. Now that Microsoft is fully committed to the AI field, the company plans to spend $50 billion in the past fiscal year to achieve its AI goals.