Between now and 2030, data centers could emit 2.5 billion tonnes of greenhouse gases globally – threefold what it would be if generative AI did not exist.
That’s according to research from investment bank Morgan Stanley published earlier this week. In “Global Data Centers: Sizing & Solving for CO2”, the company reports that decarbonization solutions will be a priority as many tech businesses aim to be carbon neutral by the end of the decade.
Gen AI, including models like ChatGPT, is capable of responding to prompts and generating its own text, images, and audio. Unsurprisingly, this uses a lot of energy. Data centers are the buildings used to store the physical computer systems that AI runs on.
What does the report say about generative AI and carbon emissions?
New facilities like these being built to house AI are a large contributor to CO2 emissions. Along with the sheer quantity of electricity required to power it all, emissions from these centers could reach 2.5 billion tonnes by the end of the 2020s.
The research also says that by 2030, data centers will be producing 600 million tonnes a year. That’s triple the 200 million tonnes that is estimated for this year.
These projections come after seeing the rapid growth and demand for the technology that gen AI has undergone in just a few years. Tech giants including Meta, Apple, and Google have all jumped on board following the success of start-ups like Anthropic and OpenAI. Investments show no signs of slowing down, as the ChatGPT maker could even be worth more than $100 billion after a new fundraising round.
Morgan Stanley’s report is targeted at those interested in investing in sustainable solutions. It highlights carbon capture and carbon dioxide removal opportunities in the industry. Investment between $15 and $45 billion could be required according to the research.
Featured image credit: Generated by Ideogram
The post Generative AI could be set to cause three times more carbon emissions from data centers appeared first on ReadWrite.
