Energy: the Next Frontier in Powering AI

Navigating the challenges and opportunities in the energy landscape driven by Artificial Intelligence

Following the rise of cloud computing, artificial intelligence (AI) is now emerging as the new focal point in the investment landscape, and for good reason. AI is driving a deep transformation that is reshaping entire sectors of the economy and radically changing the way we live, work, and interact with the world.

The rapid advancements in AI technologies, particularly in natural language processing, image recognition, and intelligent automation, are redefining workflows and optimizing value chains globally. According to a PwC study, AI could contribute up to $15.7 trillion to the global economy by 2030. As a comparison, the current US GDP is $25 trillion.

However, this revolution comes with significant challenges, particularly regarding energy consumption. The infrastructure needed to support large-scale AI, such as data centers, requires substantial energy resources. In this article, we will examine the energy requirements associated with AI and analyze their potential implications for investors.

A huge energy requirement

Data centers and their associated transmission networks have become a significant driver of global energy consumption, currently accounting for approximately 2 to 3% of total global energy usage while also emitting substantial CO2 emissions.

In the United States, the demand for electricity driven by data centers is projected to rise from 200 terawatt-hours (TWh) in 2022 to 260 TWh by 2026, representing about 6% of the nation’s total energy consumption. Furthermore, energy demands from data centers are expected to double by 2030.

A major contributor to this rising energy requirement is artificial intelligence. For instance, the data centers responsible for training and operating AI models consume vast amounts of electricity. The training of GPT-4, for example, required over 50 gigawatt-hours, approximately 0.02% of California's annual electricity generation, and was 50 times more energy-intensive than training its predecessor, GPT-3.

AI systems necessitate significant computational power, with generative AI models consuming roughly 33 times more energy to perform tasks compared to traditional task-specific software. As AI technology continues to proliferate across various industries, energy demand will inevitably increase.

Major players are actively seeking solutions

Recent developments illustrate that major players, such as Alphabet, Amazon and Microsoft, are proactively searching for solutions to address rising energy demands.

For instance, Alphabet has signed a deal with the startup Kairos Power to procure electricity from multiple Small Modular Reactors (SMR) to meet its growing electricity needs for artificial intelligence. This partnership aims to bring Kairos' first SMR online by 2030, with plans for additional reactors to be deployed by 2035. Google has committed to purchasing a total of 500 megawatts of power from six to seven reactors.

In March, Talen Energy announced a deal to sell electricity and a data center campus at its Pennsylvania nuclear power plant to Amazon Web Services (AWS). This agreement will provide AWS's data centers with up to 960 megawatts of electrical capacity, enough to power approximately one million homes.

Additionally, in September, Microsoft and Constellation Energy signed a groundbreaking power agreement to help restart a unit of the Three Mile Island nuclear plant in Pennsylvania. If successful, this would mark the first-ever restart of a nuclear facility of this nature.

Additional information available here for Microsoft, Amazon and Alphabet.

The importance and limitations of nuclear energy

These developments are particularly noteworthy as they highlight a growing reliance on nuclear energy. Nuclear power is nearly carbon-free and is often viewed as a more reliable alternative to intermittent energy sources like solar and wind, making it an attractive option for technology companies with demanding power needs and strong climate commitments.

However, several challenges remain. First, securing a stable supply of nuclear fuel is essential. Second, safety concerns may deter local communities living near reactors, despite nuclear energy being proven safe over time. Lastly, these nuclear installations require significant time to become fully operational, with most projects likely not commencing until late in the decade at the earliest.

Small Modular Reactors (SMRs) present several advantages that help mitigate these challenges. They are generally more cost-effective, require shorter construction timelines, and have a smaller physical footprint. Additionally, SMRs are designed to operate at lower pressures, which enhances their safety compared to traditional nuclear reactors.

Opportunities for investors

Investors have several avenues to capitalize on the rising energy demand associated with advancements in AI and cloud computing. It is important to note that the companies mentioned here are for illustrative purposes and may not necessarily represent sound investment choices:

• Energy Providers. Investors can consider purchasing shares in energy companies such as Vistra, Duke Energy, Constellation Energy, or NextEra Energy. These firms are well-positioned to benefit from increased energy consumption.

• Uranium Suppliers. For those interested in the uranium market, investing in uranium producers like Cameco could be an effective strategy to gain exposure to this segment.

• Energy infrastructure. Companies involved in building energy infrastructure and power plants, such as Technip Energy and BWX Technologies, are also likely to benefit from increased demand for energy solutions.

• Renewable energy and other energy sources. In addition to nuclear energy, alternative energy sources such as renewables and hydrogen are poised to play a crucial role in increasing electricity output. Companies operating in this space include Air Products, Air Liquide, and Plug Power..

• Energy storage. As the demand for electricity rises, particularly with the expansion of renewable energy, which is inherently intermittent, energy storage solutions will become increasingly vital. These technologies help balance supply and demand, ensuring a reliable power supply. ChargePoint and Tesla are both making significant strides in this area.

• Energy management solutions. Firms that provide equipment for grid management and energy efficiency, such as Schneider Electric and Eaton Corporation, represent additional investment opportunities as the need for smarter energy solutions grows.

Conclusion

As with any primary market, numerous secondary markets exist, creating a wealth of potential investment opportunities for discerning investors. While attempting to engage with every player in a particular theme often proves ineffective, identifying the specific requirements and needs within a market can uncover promising new investment themes.

Beyond investment considerations, the increased energy demand presents significant environmental challenges. At this juncture, it remains uncertain whether the efficiencies gained through AI implementation will outweigh the environmental costs, or if it will merely add to the strain on our natural resources. This complex dynamic will be an interesting topic to monitor.