Tech giants like Apple,
and
Microsoft
are working to become carbon neutral in their operations by 2030. But as the next-gen
AI
technology grows, offices and data centres are becoming more energy-hungry. As per a report published last year, by 2027 AI servers may use between 85 to 134 terawatt hours (Twh) annually – which approximately equal to what Argentina, the Netherlands and Sweden each use in a year.
To solve this problem, these companies have been pushing for
nuclear energy
as a source to power their AI tech.
“If you were to integrate large language models, GPT-style models into search engines, it’s going to cost five times as much environmentally as standard search,” Sarah Myers West, managing director of the AI Now Institute, a research group focused on the social impacts of AI, recently told CNBC.
Why nuclear energy and its limitations
Nuclear energy is seen as an alternative to
fossil fuels
to meet the rising demands of energy-guzzling data centres. One of the advantages of nuclear energy is that it doesn't emit greenhouse gases, one of the major reasons for global warming. There are two ways to generate nuclear energy: Fission and Fusion.
In fission, the nucleus of an atom splits into two or more smaller nuclei whereas in fusion, the nuclei of atoms combine while releasing energy. However, fission produces radioactive residue which is dangerous and requires proper disposal. There have also been debates on the uranium supply chain for producing nuclear energy.
Microsoft's nuclear-powered AI ambitions
Microsoft has hired Archie Manoharan as a director of nuclear technologies. She is tasked to oversee a program to develop small-scale atomic reactors – as an alternative to fossil fuels – to power the company’s datacenters up and running.
Last year, Microsoft agreed to buy power from Helion starting in 2028. Microsoft signed a deal last summer with Constellation, a top nuclear power plant operator, to add nuclear-generated electricity to its Virginia data centers.
Microsoft co-founder
Bill Gates
has also been a backer of the use of nuclear energy to keep the lights on. Recently in an interview, he was enquired about the question that he would ask a time traveller.
“How are you generating energy? Is it fusion or fission or some unexpected thing?” he said, he would ask. Gates has touted both of them as promising clean energy sources, and he co-founded nuclear energy startup TerraPower in 2006.
A similar sentiment runs with OpenAI CEO Sam Altman. He said that an energy breakthrough is necessary for future AI to run models efficiently. At the World Economic Forum's annual meeting in Davos, Altman said that future AI will consume vastly more power than people have expected.
He pointed out climate-friendly sources of energy including nuclear fusion or cheaper solar power, saying these are the way forward for AI.
In 2022, Google also invested in a nuclear fusion startup. In late 2021, Amazon founder
Jeff Bezos
was among various other investors who raised over $130 million for Canadian nuclear company General Fusion.