AI's Growing Energy Demand Drives Big Tech Toward Nuclear Power

As AI's energy needs surge, companies like Amazon are turning to nuclear power to meet these growing demands.

Amazon's Bold Move into Nuclear Power

Amazon, widely known for its expansive online marketplace and as a leader in data center operations, has taken a surprising step into nuclear power. In March, Amazon Web Services (AWS) acquired a $650 million data center campus in Pennsylvania, strategically located next to a nuclear power plant owned by Talen Energy.

This acquisition highlights Amazon's strategy to manage the immense energy demands fueled by artificial intelligence (AI). By situating its rapidly expanding AI data centers next to a reliable nuclear power source, Amazon aims to ensure the stability and scalability of its AI operations.

The Growing Energy Challenge of AI

Amazon's approach reflects a broader challenge as AI technology becomes an integral part of everyday life. AI powers a wide range of applications, from search engines to smart devices and autonomous vehicles, all of which require substantial computational power—and consequently, significant amounts of electricity.

Tech giants like Google, Apple, and Tesla continue to push the boundaries of AI, but this innovation comes at a steep energy cost. Projections indicate that by 2027, the global electricity consumption tied to AI could increase by 64%, potentially reaching 134 terawatt hours annually—comparable to the total electricity usage of nations like the Netherlands or Sweden.

Tackling AI's Insatiable Energy Needs

A pressing concern for Big Tech is how they will meet the escalating energy demands of future AI developments. According to Pew Research, 70% of Americans interact with AI at least once daily, further driving the energy consumption of data centers.

Sasha Luccioni, who leads AI and climate efforts at Hugging Face, often discusses the considerable energy usage associated with AI. She explains that while the initial training of AI models is energy-intensive, the ongoing use—known as the inference phase—where models respond to user queries, can consume even more energy due to the high volume of requests.

"For example, when a user asks AI models like ChatGPT a question, it involves sending a request to a data center, where powerful processors generate a response," Luccioni noted. "This process, though quick, uses approximately 10 times more energy than a typical Google search."

The Impact on Energy Grids

As AI-driven energy consumption continues to rise, it places additional pressure on already strained energy grids. Goldman Sachs forecasts that by 2030, the power demand from global data centers could grow by 160%, potentially accounting for 8% of total electricity demand in the United States, up from 3% in 2022.

This increasing demand is further complicated by aging infrastructure and the push toward the electrification of vehicles and manufacturing in the U.S. The Department of Energy reports that 70% of U.S. transmission lines are nearing the end of their expected 50- to 80-year life span, heightening the risk of outages and cyberattacks.

The Role of Renewables and Efficiency

While renewable energy remains a key component of Big Tech’s energy strategies, it alone may not be sufficient to meet AI’s growing energy needs. For example, in May 2024, Microsoft secured the largest corporate power purchasing agreement to date, aiming to add over 10.5 gigawatts of new renewable power capacity globally. Amazon has also positioned itself as a leader in renewable energy procurement for the fourth year running. The company's renewable energy initiatives now generate enough wind and solar power to supply the energy needs of 7.2 million U.S. homes annually.

However, the challenge with renewable energy lies in storage and timing. “The issue with renewables is that at certain times of the day, you have to also go into energy storage because you may not be using that energy at that time of the day,” pointed out Yahoo Finance reporter Ines Ferre.

In addition to expanding renewable energy sources, tech companies are investing in efficiency improvements. Google, for instance, is developing AI-specific chips, such as Tensor Processing Units (TPUs), which are designed to handle AI tasks more efficiently than traditional graphics processing units (GPUs). Nvidia’s latest Blackwell GPUs are claimed to reduce energy use and costs for AI models by up to 25 times compared to earlier versions.

The Need for Transparency and Regulatory Oversight

To effectively manage future energy demands and reduce costs, experts stress the importance of transparency and regulation. "We need more regulation, especially around transparency," said Luccioni, who is working on an AI energy star-rating project to help developers and users identify more energy-efficient models.

As tech companies continue to expand their AI capabilities, utility companies and Big Tech are expected to invest heavily—potentially $1 trillion—into AI in the coming years. While AI poses significant energy challenges, it also offers potential solutions. "AI can definitely be part of the solution," Luccioni added, noting that AI could help predict maintenance needs for infrastructure, reducing energy losses during transmission and storage.