The OpenAI app icon displayed alongside different AI apps on a smartphone.
Jonathan Raa | Nurphoto through Getty Photographs
Tech giants are turning to nuclear energy to energy the energy-intensive information facilities wanted to coach and run the large synthetic intelligence fashions behind right this moment’s generative AI functions.
Microsoft and Google are among the many corporations negotiating to purchase nuclear energy from sure U.S. suppliers to carry extra energy capability on-line for his or her information facilities.
this week Google stated it could purchase energy from Kairos Energy, a developer of small modular reactors, to assist “advance AI.”
“The grid wants the varieties of fresh, dependable vitality sources that may help the development of those applied sciences,” Michael Terrell, Google’s senior director of vitality and local weather, stated on a name with reporters Monday.
“We really feel that nuclear can play an essential position in serving to to satisfy our demand and in serving to to cleanly meet our demand in a manner that is extra across the clock.”
Google stated its first nuclear reactor from Kairos Energy can be on-line by 2030, with extra reactors working by 2035.
The tech large is not the one agency seeking to nuclear energy to comprehend its AI ambitions. final month, Microsoft signed a cope with the US vitality firm Constellation to revive an idle reactor on the Three Mile Island nuclear energy plant in Pennsylvania, whose reactor had been idle for 5 years.
The Three Mile Island plant was the location of the worst nuclear meltdown and radiation leak in US historical past in March 1979, when the lack of a water coolant by a defective valve induced the reactor to overheat.
Why are they going nuclear?
Expertise corporations are beneath stress to seek out vitality sources to energy information facilities – a key a part of the infrastructure behind fashionable cloud computing and AI functions.
Many builders lease servers outfitted with GPUs (graphics processing models) that may usually be too costly to personal outright from so-called cloud “hyperscalers” – comparable to Amazon, Microsoft and Google.
These tech giants have benefited from the wave of curiosity in generative AI functions like OpenAI’s ChatGPT. However this enhance in demand additionally led to an undesirable impact: correspondingly giant peaks within the quantity of vitality wanted.
World electrical energy consumption from the information middle, synthetic intelligence and cryptocurrency sectors is predicted to double from roughly 460 terawatt hours (TWh) in 2022 to greater than 1,000 TWh in 2026, based on a research report by the International Energy Agency.
Researchers on the College of California, Riverside, published a examine final April that discovered ChatGPT consumed 500 milliliters of water for each 10 to 50 prompts, relying on when and the place the AI ​​mannequin was deployed. That is roughly equal to the quantity of water in a normal 16-ounce bottle.
As of August, there have been greater than 200 million folks submitting inquiries to OpenAI’s standard chatbot ChatGPT each week, based on OpenAI. That is double the 100 million weekly lively customers reported by OpenAI final November.
Environmental opposition
Nuclear energy just isn’t with out controversy. Many local weather activists oppose such provides, citing their harmful environmental and security dangers, in addition to the truth that they don’t supply a real supply of renewable vitality.
“Nuclear energy is extremely costly, harmful and gradual to construct,” local weather charity Greenpeace says on its web site.
“It is typically known as ‘clear’ vitality as a result of it would not produce carbon dioxide or different greenhouse gases when electrical energy is generated, however the actuality is that it is not a believable various to renewables.”
Proponents of nuclear energy, however, say it presents a virtually carbon-free type of electrical energy and is extra dependable than renewable sources just like the solar and wind.
“If it is constructed and securitized the proper manner, I actually suppose nuclear is the long run,” Rosanne Kincaid-Smith, chief working officer of Northern Information Group, a worldwide information middle supplier, advised CNBC at a tech convention in London final week .
“Individuals are afraid of nuclear energy due to the disasters we have had up to now. However what’s subsequent, I simply do not see conventional networks being the sustained drive that continues within the improvement of AI,” added Kincaid-Smith.
Though Northern Information Group doesn’t use nuclear energy – neither is it actively exploring plans to make use of nuclear energy as an influence supply for its AI information facilities – the agency needs to “contribute to this dialog as a result of it is essential to the broader ecosystem , the broader economic system,” Kincaid-Smith advised CNBC.
— CNBC’s Pippa Stevens contributed to this report
