Source Rock Blog
The AI Boom: Why America Needs Natural Gas
March 10, 2025
America’s AI boom faces a looming obstacle: electricity generation. A single AI-powered search can consume nearly 10-times the electricity of a traditional Google search. Scale that with billions of AI queries daily, and the power needed to reliably support this emerging technology with mass-market usage starts to come into focus. Compiling the problem, AI data centers are soon expected to multiply yet utility companies are already struggling to meet today’s power demands, let alone future needs.
While nuclear, renewables, and other forms of energy can each supply a portion of the power that will be needed, they each pose inherent problems. Natural gas, therefore, is the logical source for supplying reliable, around-the-clock power that will be needed to meet AI’s demands.
Driving The News: AI’s expected demand for electricity seem insatiable and is flipping long-standing energy assumptions. Just a few years ago, analysts believed U.S. power consumption had peaked. Now, with AI data centers on the verge of rapidly multiplying, utilities, grid operators, and policymakers are racing to secure enough energy to prevent power disruptions.
- The U.S. is on track to build 80 new natural gas power plants by 2030, adding 46 gigawatts (GW) of capacity, according to Enverus.
- The U.S. Department of Energy forecasts that data centers will use between 6.7% to 12% of total U.S. electricity by 2028, up from 4.4% today.
Aaron Johnson
Vice President of
Public and Legislative Affairs
Why It Matters: Electrical reliability is at risk because AI data centers require continuous power. Grid operators, for example, are already warning of supply shortages. The North American Electric Reliability Corporation (NERC) has warned that half of the U.S. is at an elevated risk of blackouts over the next 10 years, citing surging power demand and insufficient dispatchable generation.
- Renewables alone can’t handle the load – Solar and wind can’t meet 24/7 demand. Reliable baseload power is needed. A report from the U.S. Energy Information Administration (EIA) confirms that while renewables are expanding, their variability limits their ability to provide continuous power without natural gas or other firm generation sources.
- Nuclear power is promising but slow – While nuclear plants can provide around-the-clock power, permitting and financing are far longer and more costly than for natural gas facilities. According to Goldman Sachs, no new nuclear capacity is expected to come online before 2030. Small modular reactors (SMR) are considered a long-term prospect but the technology and market are still unproven, raising the risks for scaling up to meet AI power demand.
- Natural gas is the fastest, most scalable solution – Infrastructure exists and plants using natural gas to generate electricity provide the flexibility needed to ramp up or down to meet fluctuating AI energy demands—something intermittent renewables struggle to achieve. This makes natural gas an indispensable tool in preventing grid instability while supporting rapid technological advancements.
Natural gas power plant.
Deep Dive: New AI data centers can require as much electricity as small cities and grid planners are scrambling to keep up. If the U.S. fails to expand natural gas capacity, power shortages and soaring prices could follow. Utilities are responding:
- Texas-based NRG Energy announced plans to build four new natural gas power plants by 2029 in Texas and the Mid-Atlantic that are slated to provide 1.2 gigawatts to power artificial intelligence data centers, according to Houston Public Media.
- Entergy in Louisiana and Mississippi is building several gigawatts of new gas plants to supply AI data centers for Meta and Amazon, according to OilPrice.com.
- We Energies in Wisconsin is proposing a $2 billion investment in gas infrastructure to support Microsoft’s AI hub, according to Business Insider.
What They’re Saying:
- Goldman Sachs: “We believe supporting data center driven load growth will require investment by Utilities of $50 bn in new power generation capacity. We assume a 60/40 split between gas and renewables, which we expect to drive ~3.3 bcf/d incremental natural gas demand by 2030.”
- Kinder Morgan: “This type of need demonstrates that the emphasis on renewables as the only source of power is fatally flawed in terms of meeting the real demands of the market. The primary use of these data centers is Big Tech and I believe they’re beginning to recognize the role that natural gas and nuclear must play.” Richard Kinder, executive chairman
What We’re Watching:
- Federal permitting reforms – Can Congress streamline approvals for pipelines and natural gas plants to keep up with AI demand?
- State-level restrictions – Will states opposed to natural gas like California, Massachusetts, and New York push policies that make expansion harder, risking reliability?
- Regional competitiveness – How do western states stack up to other regions across the U.S. in meeting the increased natural gas demand for AI data centers?
- Big Tech’s next moves – Publicly they push for renewables, but privately they’re locking in natural gas deals. Is reality setting in that renewables are not up to the task?
The Bottom Line: AI needs power, and lots of it. The only way to meet demand without risking blackouts or price spikes is through rapid expansion of natural gas capacity. Policymakers and energy leaders must act now—or risk watching America’s AI ambitions stall out.