Artificial intelligence (AI) is having a Bitcoin (USD-BTC) moment and the situation could rapidly deteriorate for some of the biggest companies in this rapidly evolving technology.
Bitcoin miners are energy intensive. Their thousands of computers generate excess heat that must be cooled and dissipated. The U.S. Energy Information Administration (EIA) estimates miners consume as much as 2.3% of all electricity produced in the U.S. It threatened to cause the segment to crash because of the rising costs. AI is in a similar situation but it could mushroom exponentially very fast.
In 2021, before the real explosion of AI, the EIA said data centers, one of the biggest AI users, were consuming about 2% of all the electricity produced. It noted they are “one of the most energy-intensive building types, consuming 10 to 50 times the energy per floor space of a typical commercial office building.”
And AI is an energy hog. The International Energy Agency (IEA) forecasts that data centers’ total electricity consumption could reach more than 1,000 terawatt-hours (TWh) in 2026, or “roughly equivalent to the electricity consumption of Japan.”
In the U.S. alone, Boston Consulting Group estimates AI-powered data centers will consume as much as 7.5% of electric output by 2030, triple what it was in 2022.
The energy crisis is real because there is not enough capacity available to meet projected demand. Amazon (NASDAQ:AMZN) is one of the largest data center operators for its cloud services business. CEO Andy Jassy told Fortune, “These large language models are so power hungry, and there’s just not enough energy right now.” The e-commerce giant recently purchased a nuclear-powered data center to help offset its electricity demands.
Amazon operates 125 data centers globally but just 24 in the U.S. Below are three other giant data center operators at risk because of AI’s alarming energy consumption.
Equinix (EQIX)
Equinix (NASDAQ:EQIX) is a real estate investment trust (REIT) that owns and operates 260 data centers in 71 major metropolitan areas around the globe. It understands the all-consuming nature of artificial intelligence in data centers. Equinix has been using AI-based software to reduce by 9% energy consumption at a Frankfurt data center and plans to roll it out to its other Frankfurt data centers. It was an essential response as Germany introduced an aggressive new law to increase energy efficiency.
If it wants to fulfill its expansion dreams, Equinix will need to deploy similar strategies elsewhere. The REIT wants to build out its xScale data centers for hyperscale usage. Hyperscalers are supersized cloud services providers like Amazon and Alphabet‘s (NASDAQ:GOOG)(NASDAQ:GOOGL) Google. These are the biggest drivers of AI demand as they seek out AI accelerators to boost their capacity. Google says its AI accounts for almost 15% of its data centers’ energy usage.
Without money-saving initiatives like this, Equinix could find itself pinched by growing demand of its xScale facilities.
Microsoft (MSFT)
Although it’s understood Microsoft‘s (NASDAQ:MSFT) need for data centers for its fast-growing Azure cloud service, what might not be as apparent is just how many the tech giant owns and operates itself.
According to industry site Dgtl Infra, Microsoft operates 300 physical data centers in 64 regions globally. It is in the process of adding 15 more regions to its constellation, with each region having from one to three individual data centers. Together, Microsoft’s data centers span over 75 million square feet.
Microsoft needs this web of connectivity as it is one of the most prominent companies diving head-first into AI. Beyond the scalability AI provides Azure, the tech stock has integrated it throughout all of its products and services, including Bing, Microsoft 365, Teams, Dynamics 365, OneDrive, LinkedIn, Skype, Xbox and Outlook.
Because every facet of Microsoft’s business is tied to AI, the power demands it places on the grid are excessive. It also hints at the next pain point for AI: water consumption, an equally scarce resource. Google’s data centers used 20% more water in 2022 than in 2021 while Microsoft’s water usage soared 34%.
Digital Realty Trust (DLR)
Digital Realty Trust (NYSE:DLR) is the world’s largest data center REIT with 308 locations in 25 countries. As companies continue to transition their data to the cloud, they increasingly turn to DLR to house it. Among its biggest customers are Oracle (NYSE:ORCL), IBM (NYSE:IBM) and hyperscaler Meta Platforms (NASDAQ:META).
As much of the problem as AI is, Digital Realty sees it also being part of the solution. DLR says AI can measure and reduce consumption significantly by intuitively deploying resources at appropriate times. In a 2019 pilot program using its Apollo AI platform, DLR tested conservation methods at two European facilities. It rolled it out to 14 more soon after. Combined, Digital Realty was able to save about 18 gigawatt-hours (GWh) of savings.
The REIT offers hope that even as AI drives immense demand on energy resources, it can offset that by increasing efficiency. But whether that is possible on a global scale as AI usage increases exponentially remains to be seen.
On the date of publication, Rich Duprey did not hold (either directly or indirectly) any positions in the securities mentioned in this article. The opinions expressed in this article are those of the writer, subject to the InvestorPlace.com Publishing Guidelines.