With the global popularity of artificial intelligence (AI) and its applications, a group of tech giants continue to invest in expanding data centers to train large language models, resulting in a sharp increase in electricity consumption. How to produce more clean energy to meet the ever-increasing demand for electricity has become a major obstacle to the further development of AI in the future.
According to reports, Google is collaborating with utility companies in the United States and other countries to evaluate nuclear energy as a possible source of energy for its data centers, highlighting Google's growing interest in using atomic energy to power the AI boom.
Amanda Peterson Corio, Global Head of Energy at Google Data Centers under Alphabet Inc., said, "In the highly regulated market of the United States, we do not have the opportunity to directly purchase electricity. We are working with our utility partners and power plants to explore how to introduce these new technologies into the grid - nuclear energy may be one of them
She does not rule out the possibility of Japan and other countries using nuclear energy.
According to a previous environmental report by Google, the company's global business is powered by 64% of carbon free energy, including wind and solar power. However, solar and wind energy mainly rely on available sunlight and wind, and the power supply is not stable, which has led companies including Google to consider stable clean energy sources such as nuclear energy or less used geothermal energy.
Corio said that for Google, having uninterrupted 24/7 energy is crucial when considering long-term growth.
A few days ago, Google CEO Sundar Pichai also talked about the company's measures to meet the huge energy consumption needs of AI projects: considering using nuclear power plants to power data centers, increasing investment in solar and thermal energy, etc.
This is a very ambitious goal, "he said when talking about the zero emissions target." We will still work very actively to achieve this goal. But obviously, the trajectory of AI investment has increased the difficulty of the task. We are currently considering additional investments, such as solar energy, and evaluating technologies such as small modular nuclear reactors
Tech giants favor nuclear energy
In fact, in order to meet the energy demand of AI, not only Google but also other tech giants have bet on nuclear energy. Other tech giants, including Microsoft and Amazon, also believe that nuclear energy is a stable, low-carbon energy source that can meet the soaring electricity demand of their data centers, while reducing dependence on fossil fuels and thus reducing emissions.
Recently, Microsoft announced a 20-year energy procurement agreement with the abandoned Three Mile Island nuclear power plant in the United States, where all electricity will be used to power AI data centers and cloud services. The nuclear power plant was shut down due to a core meltdown accident in 1979 and is now revived under the new name "Crane Clean Energy Center".
Meanwhile, Amazon recently purchased a nuclear powered data center in Pennsylvania. Nvidia CEO Huang Renxun recently stated that nuclear energy may be a viable option to meet the energy needs of AI.
He said that more and more data centers require renewable energy, and nuclear energy is a good choice. Huang Renxun also hinted that Nvidia may be interested in nuclear inductors as part of its broader strategy to diversify energy sources in data centers.
Nuclear energy, as an energy source and a sustainable energy source, is great. But it will not be the only energy source. We will need energy from all sources and balance energy availability, cost, and long-term sustainability, "he said.