Major tech companies are turning to nuclear power to address energy scarcity with the current grid system and to create competitive advantages for themselves.
getty
The AI strategy for big tech companies has them investing in nuclear energy to power their ambitions while simultaneously securing a resource their competitors will struggle to access. In 2025, Google, Amazon, and Meta all signed a pledge with the World Nuclear Association to triple global nuclear capacity by 2050. A big motivation for these companies to contribute to nuclear energy capacity is because of AI’s growth and need for more energy. Beyond the controversial narratives around AI and energy, there is a lesson in strategy that will further cement these companies as leaders in AI.
Tech’s Nuclear AI Strategy Addresses Energy and Sustainability
In 2024, Amazon announced multiple agreements with energy companies to build Small Modular Reactor nuclear energy projects. This is part of their sustainability goals and secures them part of the capacity from these reactors. One of the suppliers that companies like Amazon rely on are energy companies. Companies cannot keep their data centers and infrastructure running and train large models without them. These companies are expanding their data centers in the US. According to Axios, there are 3,000 new data centers already under construction or planned, in addition to the 4,000 already operational in the US. Energy suppliers have been strained for decades. In an interview with me, Hiruy Hadgu, a nuclear engineer of 15 years, stated that, “a lot of the regional transmission operators have shown that energy shortages are imminent, and there’s going to be blackouts, brownouts.” According to the Bank of America Institute, 31% of transmission and 46% of distribution infrastructure is close to or post its useful life, which is about 5 or more years old. This leaves them with two options: help expand the energy grid capacity and make data centers and chips more energy efficient. Many companies are looking to do both, but don’t want to rely on a singular point of failure if one path doesn’t work as well or materialize as quickly as the other.
Nuclear energy specifically aligns with two major initiatives within the tech industry: scalability for energy for an “AI native” future where AI is everywhere and in nearly everything, and meeting their sustainability goals. Beyond the tech industry, Hadgu shared that “the country, long before [AI], was looking at nuclear as part of the solution to address energy shortages, climate change.” SMRs produce up to 300 MW(e) per module, which is on par with fossil fuel-based energy sources. What makes them different is they tout zero carbon emissions from operations and a twenty-four seven output unlike wind or solar. Tech leaders want capacity that will scale with their AI investment that won’t be outdated in just a few years. CNBC calculated that Meta, Amazon, Alphabet, and Microsoft combined are planning to spend $320 billion on AI technologies and data centers.
How Nuclear Deals Strengthen Big Tech’s AI Strategy
This energy play is more than just capacity. When they reserve some of that capacity for themselves, that is capacity that is unavailable to its competitors and their energy needs. Their major competitors will rush to buy the rest so they aren’t left with the “leftovers” after big tech buys out what is available to reserve outside of what needs to be used for residential and other commercial areas. It’s a game of musical chairs where you don’t want to be the last one standing when the music stops.
This also prevents new entrants from competing in the same exact ways as the big tech companies already building AI at that scale. It is far too cost prohibitive for new entrants to create these agreements independently. Whoever doesn’t have a share will have to fight for the same pool of energy capacity as everyone else.
Nuclear Energy Hedges Against Uncertain Chip Efficiency
Many tech leaders have talked about a future where AI is as widespread as the internet. While experts predict that GPUs will become more efficient over time, there are still challenges. According to Nature, chip makers will need to be incentivized to make chips more efficient, and some researchers doubt if GPUs can become efficient enough to compensate for the energy consumption of the AI boom. Tech leaders cannot rely on uncertain timelines for that milestone, though they would opt to pay less for energy. Even with efficiency gains, they will still need that additional energy capacity with AI’s growth. Even if their needs never reach those heights, it helps prevent new entrants from gaining significant access to train large models within the US.
As we step into a future where the energy grid in the US is being shaped by big tech’s nuclear energy AI strategy, we can see a lesson in responding to constrained resources that also creates a competitive advantage. While the investment in nuclear energy expands the capacity of the overall energy grid, it also helps these companies fend off new entrants who will struggle to gain enough energy capacity to run data centers at the scale they do now or in the future. It also supports them with their sustainability goals. For tech giants, even something as fundamental as planning for the future infrastructure and utilities necessary for their AI investments has strategic elements to it.