While everyone’s attention is drawn to Artificial Intelligence (AI) groundbreakers such as Nvidia, OpenAI, Microsoft, and others, there are many more layers to the AI story, which is still in its infancy. There are several indirect, yet significant beneficiaries operating in the old-economy sectors, whose connection to cutting-edge tech is not necessarily obvious.
AI is Living in a Material World
Artificial Intelligence will likely become one of the biggest tech revolutions ever experienced, with the technology poised to become a multi-trillion-dollar market and change the world as we know it. Thankfully, we are not living in a Matrix-like metaverse, which means that anything digital – however advanced – still needs real-world, physical resources.
The lifeblood of AI is data, which is needed in vast amounts for the training and development of AI algorithms. As AI systems continue to grow and evolve, datasets, as well as the infrastructure supporting data storage, need to be scalable to keep up with AI’s insatiable demand for input.
This means an accelerating demand for data centers, the indispensable hubs housing the equipment – from servers and racks to cooling systems and energy – crucial for storing and managing the data. The mass adoption of AI has already spurred a data center “gold rush”, and the demand for data storage and processing capacity will soar even higher in the near future.
Powering the AI Machines
As AI’s computational power doubles every three to six months, the electricity supply struggles to keep pace. Data centers that are able to host AI-supporting GPUs (graphic processing units) consume enormous quantities of power, specifically where the training of the AI models is concerned. The power consumption related to AI deployment alone was estimated at 1 gigawatt (one GW equals one billion watts) in 2023 and is slated to grow seven- to tenfold in the next two years. If this outlook is correct, that would put an immense strain on both power grids and supply.
Of course, as we have seen multiple times in the past, free market agents usually find ways to regulate themselves for higher efficiency. Thus, there is ongoing work on energy-efficient cooling systems, which is important as cooling takes up ~40% of data center power demand. In addition, several chip producers are working on “in-memory computing” technology, which is expected to immensely speed up data access and scalability, cutting down on the amount of energy needed.
Still, no amount of cooling or efficiency improvement can solve the problem of energy demand rising at a breakneck speed. On the other hand, we can expect that in the next several years there will be many more data centers built, which will require significantly more power.
Hooked on Power
The founder and CEO of Tesla Elon Musk was ridiculed when he warned of a global power crisis by 2025 due to the exponential increase in AI computing, as well as electric vehicles (EV), also consuming substantial power. However, he is not the only one sounding the alarm.
The International Energy Agency (IEA) is also concerned about a possible shortage brought on by AI. For example, according to the agency, if Google Search is fully run by AI, it would use up to ten times more electricity than it does now. Amazon’s new data center in Virginia already burns as much energy as a major city, and the demand is rising by the week. ChatGPT alone is estimated to use as much electricity daily as a small town, and that doesn’t include video generation. Given that the AI adoption has just started, the current power supply cannot satisfy this need.
Utilities: Power Rush
Power utility companies are increasing demand forecasts by the hour, it seems, as data centers continue to gobble up every kilowatt coming their way and asking for more. The overall U.S. electricity demand growth has doubled from 2022, and it looks to be just the beginning.
The data center industry’s burgeoning demand for power is so great that the companies serving areas with high concentrations of these hubs are increasingly factoring burgeoning electricity demand outlooks into their production expansion plans. And, despite the good intentions for a transition to clean energy, these plans are fueled by fossils, mostly natural gas.
At present, renewable energy sources alone are not sufficient (or reliable and cheap enough) to provide an adequate source of power for the towering data center needs. Though nuclear power can theoretically supply all the energy needed and more, it would take years to build suitable capacity. And so, we are left with the traditional energy producers, burning aged hydrocarbons to power the most advanced tech humanity has ever known.
The Data Capital of the World
Every major change has unforeseen consequences; in the case of AI, these include the outlook for a profit surge at power utility companies serving areas with a large concentration of data centers.
The data capital of the world is Northern Virginia, representing almost 35% of the global data center capacity. Home to all major cloud providers, as well as a myriad of colocation hubs, the area enjoys ever-growing demand for data center space, which rose 20% in 2023 alone.
Most of the power for their insatiable needs is supplied by Dominion Energy (D), which, in addition to gas, oil, coal, and renewable-source facilities, also runs a nuclear power plant. Virginia is currently the only U.S. area with a significant and growing nuclear power supply, which adds to the region’s attraction since it is capable of supporting companies requiring growing energy consumption.
Dominion has already connected over 70 new centers to its power grid over the past few years, and new contracts keep piling up so fast that the company is seeking to double its production capacity. With Amazon’s planned new data center alone expected to consume as much power as the city of Seattle, there seems to be no other option.
17 Times Today’s Power Demand
Besides Virginia, the largest data center market in the U.S. has traditionally been the San Francisco Bay area, whose power is supplied primarily by Pacific Gas & Electric Company (PCG) and the non-profit Silicon Valley Power.
However, high land prices in California and power grid inefficiencies in both locations have been driving data-center construction into other markets, even before the onslaught of AI-related power-grid constraints.
Some of the fastest-growing locations are in Texas, primarily Dallas, Houston, and Austin, thanks to the Lone Star State’s lower electricity rates than the national average. The power in the area is supplied by a large number of companies, most of them privately held. However, the state is drawing local subsidiaries of large producers, such as Constellation Energy Texas, a part of Constellation Energy Corporation (CEG), and Reliant Energy, owned by NRG Energy (NRG).
Another prime data center market is Atlanta, Georgia. The Southern Company (SO) is one of the largest producers of electricity in the country, serving Southeast including Georgia. SO expects the demand for power from its customers to surge 17-fold through 2030.
A data center build-out is also well underway in Phoenix, Arizona, thanks to its proximity to major West Coast cities and the low risk of natural disasters. Most of the power in the area is provided by a non-profit Salt River Project and Arizona Public Services Company, a subsidiary of Pinnacle West Capital (PNW).
There are several more established and up-and-coming data center hubs in Oregon, North and South Carolina, Ohio, Kansas, Iowa, Indiana, and more. With the AI revolution only beginning, data centers will undoubtedly spread out across the country, demanding increased power supply – and filling the order books of electricity providers.
Conclusion: Powering the Revolution
The obvious winners of the AI revolution are those companies that are working feverishly to develop increasingly advanced technologies. However, investors could find value from those energy generation companies who are fueling the data centers that are powering the AI revolution.
For more exclusive market insights and content from TipRanks Macro & Markets research analyst Yulia Vaiman, click here.