-
AI-driven data centers are expected to increase power demand by over 160% by 2030, requiring a mix of nuclear, natural gas, renewables, and battery technology to meet round-the-clock energy needs.
-
Nuclear power offers low-carbon, reliable baseload energy, but challenges like high costs, permitting hurdles, and uranium supply constraints limit its short-term scalability.
-
Renewables will account for 40% of new capacity, but their intermittent nature necessitates backup solutions, while efficiency gains and diversified energy strategies will help offset rising emissions.
Nuclear power is set to play a crucial role in meeting the surging energy demands of data centers driven by the rapid growth of artificial intelligence. However, nuclear energy alone cannot fulfill the entire increase in power requirements. Other sources, including natural gas, renewables, and battery technology, will complement nuclear in addressing the escalating demand, according to Goldman Sachs Research.
Over the past year, several major technology companies have taken significant steps by signing contracts for new nuclear capacity, highlighting their commitment to low-carbon, reliable energy. This comes as data centers’ electricity consumption is projected to more than double by 2030. Research led by Brian Singer, Jim Schneider, and Carly Davenport forecasts the need for an additional 85-90 gigawatts (GW) of nuclear capacity globally to accommodate this growth. However, the global availability of new nuclear capacity by 2030 is expected to fall well short of this requirement, with less than 10% anticipated to come online within that timeframe.
The increasing demand for power is being driven not only by AI and broader data needs but also by a slowdown in energy efficiency improvements in data center infrastructure. Davenport notes that while data center efficiency gains once offset rising energy consumption, those advancements have decelerated, creating a “power surge” effect.
Projections indicate that data center power consumption could rise by over 160% by 2030 compared to 2023 levels. If 60% of this demand were met by natural gas and other thermal sources, global emissions would rise by approximately 215-220 million tons, equating to 0.6% of the world’s energy emissions. While renewables have the potential to address much of this increased demand, they are inherently intermittent, necessitating a reliable baseload energy source. Schneider explains that nuclear power remains the preferred option for this purpose. However, challenges such as high construction costs, permitting hurdles, specialized labor shortages, and uranium supply constraints limit its near-term scalability, leaving natural gas and renewables as more viable short-term solutions.
Despite these obstacles, the landscape for nuclear power is improving. Several governments and private sector initiatives are showing increased support for nuclear energy. In the United States alone, contracts have been signed for over 10 GW of potential new nuclear capacity, with three plants potentially coming online by 2030. Globally, countries are reevaluating nuclear as a critical energy source, with commitments to expand capacity in regions like Switzerland, the U.S., and Australia. Agreements at the 2023 COP28 summit also included a pledge to triple global nuclear capacity by 2050.
Renewable energy, meanwhile, is expected to account for approximately 40% of the new capacity required to support data center growth. While the cost of renewable energy—such as on-site solar and wind—is lower than that of natural gas, the intermittent nature of these sources presents operational challenges. Utility-scale solar plants operate for only about six hours daily on average, while wind plants run for nine, necessitating backup solutions like battery storage or grid support. Transmission costs further complicate matters, as renewable plants are often located far from urban centers, where the energy is most needed.
Thermal plants, including nuclear reactors and combined-cycle natural gas plants, offer the advantage of consistent, around-the-clock operation. As a result, data center operators are expected to rely on a mix of energy sources. Recent contracts from hyperscalers and cloud computing companies reflect this trend, encompassing investments in large-scale nuclear facilities, small modular reactors, renewable power purchase agreements, and carbon removal technologies.
Natural gas, despite its carbon footprint, remains a key component of the energy mix. Low-carbon alternatives come with a premium, and without a federal carbon pricing mechanism in the U.S., companies must navigate cost disparities. Internal carbon pricing at $100 per ton could help close the gap, making nuclear and renewable solutions more competitive with gas-fired plants.
In addition to diversifying energy sources, efficiency improvements in data center operations could further mitigate emissions. Between 2015 and 2019, workload demands tripled, yet power consumption remained stable due to advances in efficiency. While progress has slowed since 2020, continued innovation could reduce the power intensity of data centers, aligning with sustainability goals and addressing the rising energy needs of an AI-driven future.
