In the grand theater of our technology-driven economy, data centers have emerged as the indispensable backstage, tirelessly storing, processing, and disseminating the vast oceans of data that fuel our digital lives. With over 10,000 dedicated data centers worldwide, nearly a quarter reside in the United States, a testament to our central role in this burgeoning sector. As artificial intelligence (AI) takes center stage, the pressing question has shifted from how to harness this transformative technology to a more fundamental concern: how to power it.
Not long ago, the world fretted over the energy demands of cryptocurrency mining, which would require the relentless output of twenty-eight coal-fired power plants to satiate. That concern, it seems, is now a relic of the past. Today, AI is sparking a renewed focus on Natural Gas and a nuclear renaissance, drawing surpisingly little opposition from the public or government. By 2027, AI is projected to consume about 2% of total U.S. electricity, with a staggering growth rate of 45%. Major players like Amazon, Microsoft, Meta, and Google have announced plans to secure nuclear energy to meet their future power demands, yet projections indicate that the nuclear capacity needed to satisfy this appetite will fall short this decade.
Complicating matters, Russia currently controls 65% to 80% of the processed uranium fueling U.S. nuclear plants. This geopolitical reality means that until domestic and Canadian uranium mines can ramp up production, processing and disposal, natural gas will serve as the bridge to fill the impending energy gap. Natural gas, which has been pivotal in reducing U.S. carbon emissions, may soon see its long-standing supply surplus dwindle as demand rises.
The Trump Administration’s promised reversal of the January 2024 energy moratoriums will speed up permits for gas pipelines and liquefied natural gas (LNG) terminals. If unleashed, natural gas could transition from surplus to shortage by the decade’s end.
The forecasts for power demand are nothing short of exponential. By 2030, the electricity needs of data centers are expected to triple current capacities. Hyperscalers—those giants like Meta, Google, Amazon, and Microsoft—are anticipated to fulfill about 70% of this demand. As they expand, they will rely heavily on natural gas, which is projected to account for 60% of this growth while laying the groundwork for a nuclear future. Utilities such as Georgia Power and Southern Company have noted that their load growth projections are now 17 times higher than just two years ago. In North and South Carolina, Duke Energy reports eightfold increases in load projections, necessitating the construction of numerous new power plants. The electricity demand for data centers in the U.S. is set to increase by about 400 terawatt-hours, growing at an annual rate of 23%. McKinsey research paints a rosy picture for generative AI, estimating it could create between $2.6 trillion and $4.4 trillion in economic value over the next five years.
This demand outstrips historical capacity expansions, which have hovered around 3% annually. Over the last 10 years, natural gas demand has grown by 50%, pipeline capacity has grown by 25%, but power storage capacity has not grown at all. Newer data centers now consume as much electricity as 100,000 homes, and the total market is doubling every three years.
As we look toward 2030, AI data centers are projected to add 323 terawatt-hours of electricity demand—an increase that would require the equivalent of seven New York City’s worth of electricity. During this time, the share of total U.S. electricity consumption attributed to data centers will quadruple, rising from 2% to 8%. While natural gas and solar energy are poised to scale up to meet this demand, the world is also preparing for a nuclear renaissance to power our AI future in the 2030’s and beyond.
The infamous Three Mile Island facility is expected to restart by 2028, largely due to Microsoft’s intervention, while Amazon commits $500 million to nuclear power generation. Blackstone is i7 billion venture to build data centers and small modular nuclear reactors (SMRs) that promise cheaper, safer, and faster deployment than their atomic predecessors.
As the political landscape shifts in 2025, aggressive deregulation is their pledge. As the world races for AI leadership at an accellerating rate, it’s important that the lengthening lead times in building data centers is reversed by removing Government barriers.
The vast majority of the stock market has lagged AI related mega tech stocks since the 2022 Bull market began. Everyone knows that Nvidia has been at the vanguard of this information technology wave. In 2023 Nvidia was the top SP 500 Index performer, yet despite a stellar 200% gain so far in 2024, it lags a conservative Utility stock. Vistra Energy, with 81% of its energy generated from Natural Gas, is the best-performing stock on the S&P 500 Index so far in 2024 with gains of 300% this year.
Emerging players like Vertiv, known for its liquid immersion cooling systems for data centers, have also seen remarkable gains—70% in just two months and 165% for the year. Other high-risk stocks, backed by Nvidia, such as SoundHound and Serve Robotics, are attracting attention for their innovative AI applications, despite their speculative nature.
As we look ahead, natural gas will continue to play a crucial role in powering AI, with companies like Vistra and Constellation Energy poised to benefit. AI-related firms, including Nvidia, Vertiv, Oracle, Meta, Amazon, and Microsoft, are likely to remain at the forefront of this investment wave. Caution is warranted regarding SMR nuclear stocks, as they remain speculative with no material sales expected this decade. However, the demand for AI-driven solutions is undeniable, ensuring that investment dollars will continue to flow toward this promising frontier.
