Big tech companies are turning to nuclear power to fuel their energy-intensive AI ambitions

Tech giants are turning to nuclear power to power the energy-intensive data centers needed to train and run the massive AI models behind today's generative AI applications.

Microsoft Google is among the companies that have agreed to deals to purchase nuclear power from select suppliers in the United States to provide additional online power capacity for its data centers.

this week, Google It said it would buy power from Kairos Power, a developer of small modular reactors, to help “advance artificial intelligence.”

“The grid needs these types of clean, reliable energy sources that can support the construction of these technologies,” Michael Terrill, senior director of energy and climate at Google, said on a call with reporters on Monday.

“We feel that nuclear power can play an important role in helping to meet our demand, and helping to meet our demand in a clean, around-the-clock way.”

Google said its first nuclear reactor from Kairos Power will be operational by 2030, with more reactors operating until 2035.

The tech giant isn't the only company looking to nuclear power to fulfill its AI ambitions. last month, Microsoft The United States signed an agreement with the American energy company Constellation to revive a mothballed reactor at the Three Mile Island Nuclear Power Plant in Pennsylvania, whose reactor has been idle for five years.

The Three Mile Island plant was the site of the deadliest nuclear meltdown and radiation leak in U.S. history in March 1979, when the loss of water coolant through a faulty valve caused the reactor to overheat.

Technology companies are under pressure to find the energy sources to power data centers, an essential part of the infrastructure behind modern cloud computing and artificial intelligence applications.

Many developers rent servers equipped with graphics processing units (GPUs), which are usually too expensive to own outright, from so-called cloud “hyperscalers” – such as Amazon, Microsoft and Google.

These tech giants have benefited from increased interest in generative AI applications like OpenAI's ChatGPT. But this increase in demand also led to an unintended effect: significant increases in the amount of energy required.

Global electricity consumption from data centres, artificial intelligence and the cryptocurrency sector is expected to double from an estimated 460 terawatt-hours in 2022 to more than 1,000 terawatt-hours in 2026, according to a research report from the International Energy Agency.

Researchers at the University of California, Riverside, published a study in April last year that found that ChatGPT consumes 500 milliliters of water for every 10 to 50 requests, depending on when and where the AI ​​model is deployed. This is roughly equivalent to the amount of water in a standard 16-ounce bottle.

As of August, there were more than 200 million people submitting questions on OpenAI's popular chatbot ChatGPT each week, according to OpenAI. That's double the 100 million weekly active users that OpenAI reported last November.

Nuclear power is not without controversy. Many climate activists oppose such supplies, citing their serious environmental and safety risks, and the fact that they do not provide a true source of renewable energy.

“Nuclear power is incredibly expensive, dangerous, and slow to build,” climate charity Greenpeace says on its website.

“It is often referred to as ‘clean’ energy because it does not produce carbon dioxide or other greenhouse gases when generating electricity, but the reality is that it is not a reasonable alternative to renewable energy sources.”

On the other hand, proponents of nuclear power say it provides a nearly carbon-free form of electricity and is more reliable than renewable sources such as solar and wind.

“If it's built and secured in the right way, I think nuclear power is the future,” Rosanne Kincaid Smith, chief operating officer of Northern Data Group, a global data center provider, told CNBC at a technology conference in London last week. .

“People are afraid of nuclear power because of the disasters we've seen in the past,” Kincaid-Smith added. “But what's to come, I don't see traditional grids being a continuing sustainable force in the development of artificial intelligence.”

Although Northern Data Group doesn't use nuclear power — nor is it actively exploring plans to use nuclear power as a power source for its AI data centers — the company wants to “contribute to that conversation because it's important to the broader, broader ecosystem.” The economy,” Kincaid Smith told CNBC.

— CNBC's Pippa Stevens contributed to this report