Last updated on April 23, 2025
It’s not a surprise that we’re seeing billions poured into AI technology due to the demanding growth of AI-oriented platforms such as ChatGPT. Still, there is an underlying issue with this technology’s widespread use.
With rising demand for AI software, leading cloud service providers (CSPs) will demand more energy for their data centers. Some CSPs including Fortune 500 companies Microsoft, Amazon, and Google are jumping on alternative clean energies to meet this demand. Historically, many of these data centers were strategically placed near hydroelectric sources and solar farms to generate the power necessary. These alternative sources, however, cannot meet the demand for AI processing, which makes these up-and-coming energy sources inadequate. So what’s the viable alternative that blue chip companies are betting billions on?
Nuclear energy.
For context on the power needed, a Google search takes about 0.0003 kilowatt-hours of energy—roughly equivalent to powering a 60-watt light bulb for around 20 seconds. Using an AI prompt takes 10 times more energy according to NPR, and researchers project that data centers will go from consuming 2-3% to roughly 9% of U.S. annual energy. With almost triple the energy demand, solar power has only been able to generate 2% of the country’s total energy output, per the U.S. Energy Information Administration. In comparison, the total electricity generation from nuclear power was nearly one-fifth of total production.
Demand for AI will continue to expand, and thus, energy consumption will grow in the following decades. With nuclear plants able to generate the hundreds of megawatts required, atomic energy is the alternative needed in the foreseeable decades.
Continuous reliance on fossil fuels is unlikely with companies making environmental promises to their investors, and state legislatures like California mandating renewable energy usage in the foreseeable future.
Some sustainable energy sources including solar panels were never capable of generating the hundreds of megawatts needed to power small cities, and now alternatives must solve the AI software issue. Breakthroughs in certain developments including battery storage would make these clean energy sources economically feasible. For now, implementing solar and wind power as an energy source for data centers is not viable.
Being a third of the size compared to standard nuclear power plants, small nuclear or modular reactors have recently been the most sought-after alternative by companies. Investing billions in modular reactors is the best choice for companies as it allows them to remain carbon neutral and meet the energy demand.
Hence, we are seeing partnerships between CSPs and companies that specialize in building small modular reactors (SMRs) like Constellation Energy. In a partnership with an SMR company, Kairos Power, Google intends to build seven different reactors — able to generate a total of 500 megawatts of energy. Similarly, Amazon Web Services has begun constructing three reactors that cost hundreds of millions each year.
Amidst the ambitious plans, America’s nuclear energy sector has historically remained stagnant. The Chernobyl atrocity in 1986 and the U.S. Three Mile Island meltdowns in Pennsylvania have made this energy source unappealing — largely because of fear. It is hard to believe that any dismissive public opinion will halt the billions already invested into modular reactors. In fact, polling data shows the changing narrative around this technology.
Since 2020, although public support for solar and wind power has declined, nuclear power’s favoritism has grown 13% according to the Pew Research Center. When CSPs incorporate reactor plants near neighboring communities, public opinion will only continue to grow.
In a sense, AI has impacted the lives of millions with its unprecedented abilities. It has changed our relationships with social media, schooling, and other aspects of our livelihoods. The change in our renewable energy source has been only one byproduct in AI’s transformative nature.
In the long run, modular reactors may not be the only promising revelation. Nuclear fusion, which mimics the atomic reaction in the sun, would provide an even cleaner energy source than the current nuclear reactors. The emerging technology promises limitless carbon-free power without the dangerous nuclear waste. Currently, Lockheed Martin hopes to have a working prototype in five years.
AI will have long been developed when nuclear fusion becomes widespread, so for now, modular reactors will suffice.
Be First to Comment