Two major trends have collided in President Trump’s return to the White House. In his first day in office, Trump declared a “national energy emergency“ which aims to reward Big Oil donors that backed his campaign by accelerating the drilling of fossil fuels in the U.S. at the expense of green energy initiatives favored by climate scientists and the prior administration.
Meanwhile, the White House announced the $500 billion Stargate Initiative, which aims to provide widespread infrastructure for artificial intelligence (AI) efforts in the U.S.
Both sides — the “drill, baby, drill!” crowd on one side and environmentalists on the other — likely don’t share much common ground. But they both seem to agree that another generational trend — the rise of AI — is forcing us to find more energy than we could have imagined just a few years ago.
As data centers require more power, particularly for generative AI (GenAI) and large language models (LLMs), the AI revolution is going to rely on a wave of new electricity coming online.
What will these energy sources look like? And how does the rise of AI potentially affect the fight against climate change?
AI and Energy: A Symbiotic Relationship
Before diving into their energy usage, it’s worth breaking down the two primary types of AI industry players: semiconductor manufacturers on one side and model providers on the other. Manufacturers, including industry behemoths Nvidia, Advanced Micro Devices, and Taiwan Semiconductor build graphics processing units (GPUs) that provide the computational power used in LLMs and GenAI applications.
The model providers include pure-play AI companies such as OpenAI (creator of ChatGPT) and Anthropic (creator of Claude), along with tech industry titans Alphabet, Microsoft, Amazon , and Apple, among many others. It’s estimated that the AI model providers will invest upwards of $200 billion annually on GPUs and other AI infrastructure by 2028. While China’s DeepSeek at first seemed like it might disrupt the spending spree, the AI “space race” seems to have remained on track or even accelerated.
How Much Energy Does AI Require?
While asking ChatGPT a simple question may seem straightforward, the amount of spending by AI companies shows there’s a massive amount of computational power required to answer even basic requests. Providing the energy to fuel AI models has become “mission-critical” to maintaining advances in AI.
More powerful GPUs require ever-increasing amounts of energy. Nvidia’s famed Blackwell chip — which claims to offer five times better performance than its predecessor chip — requires 1.2 kilowatts (KW) of electricity per hour, equivalent to the amount of electricity an average U.S. household consumes in one day. These numbers add up.
In 2022, the International Energy Agency (IEA) estimated that data centers consumed 460 terawatt-hours (TWh), or about 2% of all the energy consumed in the world. (A terawatt hour equals one trillion KWh.) Amidst the AI boom, the IEA expects that number to more than double by 2026, likely reaching 1,000 TWh. While that number may sound abstract, it equals the amount of total energy consumed by Japan in one year. By 2026, AI will require as much electricity as the world’s fourth-largest economy.
New Energy Sources, New Investment Opportunities
The demands for more power for data centers have already put significant stress on aging global energy infrastructure. To keep AI progress marching forward, we’ll need more electricity, and soon.
According to the U.S. Energy Information Agency (EIA), fossil fuels comprised nearly 85% of the U.S. energy supply in 2023, with the remainder made up of renewable and nuclear sources. While natural gas and oil will likely remain cornerstones of the U.S. energy supply (particularly under a GOP government), renewable and nuclear energy sources have emerged as sustainable ways to power AI’s nearly insatiable demands for electricity.
For instance, in September 2024, Microsoft announced that it planned to open a formerly mothballed nuclear reactor at Three Mile Island in a partnership with utility company Constellation Energy. The fact that the companies chose Three Mile Island — a Pennsylvania nuclear plant infamous for a near-meltdown in 1979 — shows that proliferating energy demands are leading to some creative thinking.
Along with nuclear, renewable sources — especially wind and solar — could also play a growing role in providing power for AI. In fact, the EIA predicted in early 2024 that solar and wind would be the fastest-growing energy sources over the next two years, with U.S. solar power generation projected to increase 75% from 163 billion kWh in 2023 to 286 billion kWh by the end of 2025, while wind power generation was set to grow 11% from 430 billion kWh in 2023 to 476 billion kWh in 2025. Despite these gains, Trump’s actions to rescind climate financing initiatives put in place by the Biden Administration could affect this growth trajectory.
Finally, from an investment perspective, the AI boom has led to some surprising stock market dynamics driven by data center power needs. Historically known for defensive properties and favored in challenging market conditions, utility company stocks (+24.7%) outperformed the broader S&P 500 (+23.1%) in 2024 on strong demand for power. As the thirst for power to fuel the AI boom continues, once-staid utility companies could emerge as growth opportunities for investors in the years ahead.
AI and Sustainability
AI has fundamentally transformed how many people interact with information, with services like ChatGPT poised to answer any question we can think to ask it (even if some of those answers are the result of AI hallucinations). But few people likely ask themselves how much power and water ChatGPT requires to answer those questions.
According to recent research, each ChatGPT prompt on average requires 2.9 kWh of energy, ten times the amount of a single Google search and enough power to illuminate a 60-watt light bulb for 17 seconds. Cumulatively, ChatGPT usage is annually expected to account for 226.8 GWh of energy usage, enough to power approximately 20,000 U.S. homes for an entire year and provide enough electricity to power the country of Belgium for a day. And as a note, this only covers ChatGPT and does not account for the many other LLMs and GenAI models in use today.
Along with consuming energy, answering AI prompts also requires significant amounts of water. A 2024 study found that ChatGPT requires two liters of water to answer between 10 and 50 questions. Again, these numbers add up. Research from MIT’s Climate Project argued that the unconstrained growth of AI could place significant strains on municipal water supplies, as vast amounts of water are needed to cool data centers, generate thermoelectric and hydroelectric power, and facilitate manufacturing of AI chips. In the UK, the issue of unregulated and opaque water usage by data centers has become a hotly contested political topic.
Looking Ahead: AI and Climate Change
The AI revolution is already here and will likely continue to change our world in ways that we don’t fully understand yet. At the same time, the climate crisis continues to grow in severity, though there’s still hope to combat the worst effects of a warming planet. For now, it’s important to realize that these foundational disruptions are bound tightly together and will likely influence each other over time. In fact, AI could actually help fight climate change, even as it may also contribute to global warming and resource depletion.
Like all revolutions, AI is likely to produce winners and losers. Just remember that we only have one planet, and even the smartest algorithms won’t be able to create a new one.
For more insights and guidance on navigating the evolving landscape of AI, LLMs, energy production sustainable investing and other related issues, stay tuned to our blog for future updates and expert analyses.
And help us build a more sustainable and prosperous world through responsible investment practices by becoming a member of the Advance ESG community.
It’s free to join and there are no future financial obligations. Together, we can make a difference in safeguarding our planet for future generations.
