As of March 13, 2025, the rapid expansion of artificial intelligence (AI) and its supporting infrastructure, particularly data centers, is reshaping US electricity consumption. This analysis explores how this increased demand will create a significant boom for the natural gas industry over the next eight years, focusing on the role of natural gas as a dispatchable energy source. The analysis incorporates data points from recent projections, including AI-related electricity consumption estimates and natural gas use, to provide a detailed overview.
Growing Electricity Demand from AI and Data Centers
AI-driven data centers require vast amounts of electricity to operate continuously, driven by the computational intensity of AI workloads such as training large language models. Recent estimates indicate that AI data centers and servers consumed approximately 116 TWh in 2020, 136 TWh in 2021, 156 TWh in 2022, 176 TWh in 2023, and 196 TWh in 2024 in the US, based on total data center consumption with AI as a major driver. Projections for AI-related industry electricity use, factoring in recent announcements like the Stargate Project—a $500 billion investment in AI data centers starting in 2025—suggest consumption will range from 108 TWh in 2025 to 473 TWh in 2033. This growth reflects the rapid expansion of AI applications and new data center developments, potentially adding up to 200 TWh by 2033 from projects like Stargate.
Natural Gas: The Dispatchable Energy Source
Natural gas plays a critical role in US electricity generation, accounting for about 40% of total electricity generation in recent years, with 42% in 2023 according to the US Energy Information Administration (EIA) (EIA Electricity Generation Data). Its dispatchable nature—meaning it can be turned on or off quickly to meet demand—makes it ideal for powering data centers that require a steady and reliable supply. Unlike renewable sources like wind and solar, which are intermittent and non-dispatchable due to weather dependency, natural gas provides grid stability, essential for meeting sudden increases in data center demand. In 2023, natural gas consumption for electricity generation averaged 35.4 bcf/d, representing 40% of total US natural gas consumption of 89.4 bcf/d (EIA).
Conversion Factors and Consumption Estimates
To estimate the additional natural gas needed for AI-related electricity demand, we calculated the conversion factor based on 2023 data. Total electricity generated from natural gas was approximately 1,692 TWh, requiring about 12.931 Tcf (or 12,931 bcf) annually, translating to a rate of approximately 7.64 bcf per TWh, assuming an efficiency of around 45%. This means for each TWh of AI-related electricity demand met by natural gas, about 7.64 bcf of natural gas is needed annually. Assuming natural gas maintains a 40% share in meeting this demand, we can project the additional consumption:
- For 2025, with AI-related demand at 108 TWh, natural gas contribution is 43.2 TWh, requiring approximately 330.288 bcf annually, or 0.905 bcf/d.
- For 2033, with demand at 473 TWh, natural gas contribution is 189.2 TWh, requiring approximately 1,446.528 bcf annually, or 3.963 bcf/d.
These calculations assume total US electricity generation projections from EIA’s Annual Energy Outlook 2023 (AEO2023), estimating generation at 4,145 TWh in 2025, increasing to 4,496.4 TWh in 2033, with natural gas’s share slightly decreasing from 42.2% to around 40% (EIA Annual Energy Outlook).
Projected Increase in Natural Gas Consumption
The additional natural gas consumption due to AI-related demand ranges from 0.905 bcf/d in 2025 to 3.963 bcf/d in 2033, representing a significant increase for the natural gas industry. Current total natural gas consumption is 90.4 bcf/d in 2024, according to EIA. By 2033, without AI-related demand, projections suggest consumption could reach around 98.6 bcf/d (interpolating from AEO2023, total consumption at 36 Tcf annually, or 98.6 bcf/d). With AI-related demand, total consumption could reach 102-104 bcf/d, a 4-6% increase, driven by data center needs.
AI Demand Projections:
| Year | AI-Related Electricity Demand (TWh) | Natural Gas Share (TWh) | Additional Natural Gas Consumption (bcf/year) | Daily Increase (bcf/d) |
|---|---|---|---|---|
| 2025 | 108 | 43.2 | 330.288 | 0.905 |
| 2026 | 156 | 62.4 | 477.776 | 1.31 |
| 2027 | 205 | 82 | 626.48 | 1.716 |
| 2028 | 254 | 101.6 | 776.384 | 2.127 |
| 2029 | 303 | 121.2 | 926.248 | 2.538 |
| 2030 | 352 | 140.8 | 1,075.152 | 2.946 |
| 2031 | 400 | 160 | 1,222.4 | 3.349 |
| 2032 | 449 | 179.6 | 1,373.104 | 3.762 |
| 2033 | 473 | 189.2 | 1,446.528 | 3.963 |
This increase, particularly the nearly 4 bcf/d by 2033, represents about 4.4% of current consumption, likely spurring production and infrastructure investments. An unexpected detail is how natural gas producers, like those in the Permian Basin, are directly engaging with data center operators, potentially easing price slumps and fostering new partnerships.
Demand and Supply Projections
To illustrate the scale, consider the following table comparing total US electricity consumption and natural gas consumption for AI-related demand:
| Year | Total US Electricity Generation (TWh) | AI-Related Electricity Demand (TWh) | Natural Gas Consumption for AI (bcf/d) |
|---|---|---|---|
| 2025 | 4,145 | 108 | 0.905 |
| 2030 | 4,374 | 352 | 2.946 |
| 2033 | 4,496.4 | 473 | 3.963 |
This table highlights AI’s growing share, potentially increasing natural gas consumption by nearly 4 bcf/d by 2033, reinforcing its role in the energy mix.
The Stargate Project
Recent announcements, particularly the Stargate Project, a joint venture involving OpenAI, SoftBank, Oracle, and others, plan to invest up to $500 billion over four years starting in 2025 to build AI data centers in the US. The project begins with a $100 billion initial investment, with the first 1 million-square-foot data center already under construction in Texas. To estimate its impact, we assume Stargate’s data centers are entirely AI-related and will scale linearly from 25 TWh in 2025 to 212 TWh by 2033, reflecting a phased build-out over eight years:
- 2025: 25 TWh
- 2026: 55 TWh
- 2027: 75 TWh
- 2028: 110 TWh
- 2029: 137 TWh
- 2030: 161 TWh
- 2031: 177 TWh
- 2032: 190 TWh
- 2033: 212 TWh
Conclusion
In summary, research suggests AI and related industries will increase US electricity use significantly, from 108 TWh in 2025 to 473 TWh in 2033, driving natural gas demand. It seems likely that natural gas, as a dispatchable source, will meet much of this demand, with consumption potentially adding 0.9 to 4 bcf/d by 2033, leading to a boom for the industry. The evidence leans toward increased production and infrastructure investments, though controversy persists over renewable competition and environmental regulations. This shift underscores natural gas’s critical role in supporting AI’s energy needs over the next eight years.
Talk to Us
Contact our Commodity Brokers to explore how investing in Natural Gas can work for you. Visit Paradigm Futures to learn more.
Full Disclaimer
The risk of loss in trading futures and/or options is substantial, and each investor and/or trader must consider whether this is a suitable investment. Past performance is not indicative of future results. Trading advice is based on information taken from trades, statistical services, and other sources that Paradigm Futures believes to be reliable. We do not guarantee that such information is accurate or complete, and it should not be relied upon as such. Trading advice reflects our good faith judgment at a specific time and is subject to change without notice. There is no guarantee that the advice given will result in profitable trades.
