Microsoft has announced an energy deal to reopen the Three Mile Island nuclear power plant on the Susquehanna River near Harrisburg, Pennsylvania.
Constellation Energy, the present owner of the plant, is now tasked with bringing Unit 1 back online for Microsoft, requiring an investment of $1.6 billion to revive the reactor by 2028.
While details are usually not yet known, Microsoft is claimed to have offered to purchase the plant's production for 20 consecutive years.
Three Mile Island is best generally known as the positioning of the worst nuclear accident in US history. In 1979, a partial meltdown occurred in one among the reactors, which caused fear and mistrust of nuclear power among the many population.
Reactor Unit 2 of the ability plant, where a core meltdown occurred, was permanently closed, while Unit 1 remained in operation until its shutdown in 2019 resulting from competition from cheaper natural gas.
According to Microsoft, the deal can be resulting from the corporate's commitment to zero carbon emissions by 2030. Nuclear energy is a carbon-free energy source, but there may be ongoing controversy over the disposal of radioactive waste.
Joseph Dominguez, CEO of Constellation Energy, commented positively on the move: “This plant should never have been allowed to shut. It will produce as much clean energy as all of the renewables (wind and solar) inbuilt Pennsylvania over the past 30 years.”
Constellation Energy said the ability plant would require “significant investments”, including upgrading and refurbishing the “turbine, generator, foremost transformer, and cooling and control systems”.
The increasing performance requirements of AI
Microsoft's decision to make use of nuclear energy once more demonstrates the big energy requirements of artificial intelligence and the supporting data center technology.
The company has expanded its data centers around the globe, lots of which concentrate on supporting AI workloads, including training and deploying models that require massive amounts of computing power.
Training large AI models can devour hundreds of megawatt hours (MWh) of electricity.
For example, in accordance with some sources, training OpenAI's GPT-3 required over 1,200 MWh, which could power tens of hundreds of homes for a day.
Today, lots of, if not hundreds, of powerful AI models are trained concurrently. AI models require power not only during training, but in addition in every day operation.
This increase in energy demand resulting from AI is an element of a broader trend. The International Energy Agency (IEA) estimates that data centers currently 1.3% of worldwide electricity Consumption, and this can increase significantly, with AI infrastructure accounting for a big a part of this increase.
By 2030, data centers could have as much as 8% of worldwide electricitywhich puts further strain on already overburdened energy grids resulting from increasing dependence on digital services and electric vehicles.
Coal and nuclear energy to fill the gap
While the concentrate on nuclear energy underscores the technology industry's need for low-carbon alternatives, AI's energy needs are remarkably respiration latest life into coal.
According to a Bloomberg Report from the start of the yrThe rapid expansion of information centers is delaying the closure of coal-fired power plants within the United States, thereby resisting the push toward cleaner energy sources.
In areas like Kansas City, for instance, the development of information centers and electric vehicle battery factories has forced utilities to desert plans to shut coal-fired power plants.
Microsoft's decision to power its AI activities with nuclear energy brings the broader discussion concerning the sustainability of AI into focus.
As the technology industry grows faster than energy supply, progressive solutions are needed to bridge the gap between demand and production. For example, OpenAI has actively invested in Helion, a nuclear fusion project that is predicted to return online soon.
Sam Altman, CEO of OpenAI, said at X: “If Helion works, it is just not only a possible way out of the climate crisis, but in addition a path to a significantly higher quality of life. I even have enjoyed being involved for the past 7 years and look ahead to investing more.”
Despite its controversies, nuclear energy offers a reputable solution to the energy needs of artificial intelligence, especially in regions that struggle to completely transition to renewable energy.
But the stakes are high. Building and maintaining nuclear power plants still requires enormous resources, and managing nuclear waste is a challenge. Many will see this as a trivialisation of strategies to decarbonise and promote renewable energy.
We need to say that the deal between Microsoft and Constellation Energy remains to be in its early stages.
Yet exclusive, private deals like this are extremely rare, demonstrating how much power within the AI industry is determined by power within the literal sense.
