In the era of artificial intelligence, data centers have become the silent engines powering our digital world, but their relentless thirst for electricity is sparking questions about sustainability and the future of global energy demands.

I didn’t write that sentence, ChatGPT did.

Artificial intelligence. Depending on what you believe, it’s either going to save us or destroy us. In the energy sector, it is certainly a heavily discussed development because it poses unique challenges and opportunities for the reliability and resilience of the electric grid system.

AI-driven data centers run high-performance computing tasks like machine learning, natural language processing, and image recognition. Those processes require massive amounts of electricity. Recent estimates suggest that data centers already account for around 1–2% of global electricity consumption, and with the surge in AI, demand is expected to rise.

In the U.S., AI demand alone could add up to 20% of our power needs over the next decade. At a time when we barely have enough power to serve the need today and political and regulatory headwinds have greatly slowed down the development of new power plants, that kind of growth in demand for electricity seems daunting.

AI data centers tend to operate continuously, leading to a higher base load on the grid. This constant demand can strain existing infrastructure, particularly in regions without robust power supply systems. They can also create unique challenges when we have moved away from 24/7/365 power plants to intermittent power producers, like wind and solar.

The bellwether for this tension is Northern Virginia, which houses nearly 50% of all U.S. data centers. Today, those facilities demand a whopping 2.8 GW of power, and that is expected to grow to over 20 GW in the next 15 years. To put that in perspective, Michigan’s total electric load is 20 GW.

While Michigan is by no means a data center haven, we do anticipate data center growth here as well. As we prepare for that growth, projects like the Palisades Nuclear Power Plant restart and the future development of small modular nuclear reactors on that site are vital to ensuring we have the energy we need to serve those data centers.

While AI causes challenges, it also poses solutions that can be used to manage the rising demand for electricity. AI can help us better predict energy demand. By analyzing real-time data, AI models can forecast electricity load patterns and respond proactively, minimizing blackouts and reducing the need for costly, unplanned maintenance.

AI can also help consumers optimize their own energy use. Smart meters and connected home devices use machine learning to identify energy usage patterns, allowing households and businesses to reduce consumption and save costs without sacrificing comfort.

Back to the question we started with—will AI destroy us or save us? Probably neither. But it is an inevitable part of the future and we must plan today for that future. Or at least that’s what ChatGPT told me when I asked it.