“A game-changing moment” — How does Microsoft’s new nuclear alliance fuel its carbon-negative ambitions?

by

In simpler terms, Generative AI requires a significant amount of energy due to its increasing need for larger data centers. This increased demand poses a challenge to current power supply systems. Microsoft proposes that the solution to this escalating energy issue could be nuclear power, and they have affiliated with the World Nuclear Association (WNA), which Dr. Sama Bilbao y León, WNA’s Director General, describes as an important shift for our industry.

When a globally recognized pioneer in technology deems nuclear energy crucial for their carbon-neutral future, it sends a potent message to financial markets, lawmakers, and influential figures across industries worldwide. This collaboration aims to expedite the widespread implementation of nuclear power at the necessary pace to achieve both climate targets and the expanding energy requirements from data centers.

Director General of WNA Dr. Sama Bilbao y León

According to the press release by WNA, as reported by TechRadar, Microsoft has shown an interest in working together on several significant subjects.

  • Advanced Nuclear Technologies — Microsoft will support the WNR and its other members in the “development and deployment of small modular reactors (SMRs) and next-generation reactors.” This included fusion energy.
  • Regulatory Efficiency — Microsoft will advocate for “streamlined licensing processes” that reduce startup times while maintaining proper security protocols.
  • Supply Chain Resilience — Microsoft will aid in bolstering the global supply chains required for nuclear energy.

Under the guidance of Dr. Melissa Lott, the Energy Technology team within our company is managing the recent collaboration with WNR. According to Dr. Lott, this partnership signifies a pivotal juncture in our industry’s evolution as we strive towards achieving our objective of carbon-free energy production.

By integrating Microsoft’s advanced technologies with the established success of the nuclear sector in providing dependable, zero-emission base power, we establish a solid platform for groundbreaking advancements in the field of deploying carbon-free energy technology.

Dr. Melissa Lott, Microsoft Energy Technology

Joining the Partnership on Artificial Intelligence (WNA) appears to be a logical progression for Microsoft, as many leading AI and AI-associated companies recognize that AI’s growing energy demands are adversely affecting climate goals and causing power grid failures.

It’s worth noting that Microsoft has shown an interest in nuclear energy as a potential solution to its issues before. In fact, back in 2023, when Microsoft’s AI division was still in its infancy, they formed a partnership with Helion, a company specializing in nuclear fusion technology. Helion aims to start producing power by the year 2028.

In the year 2024, Microsoft entered into an agreement spanning over two decades with Constellation Energy for the revitalization of the Three Mile Island Unit 1 nuclear power plant. The aim is to generate approximately 835 megawatts of energy without carbon emissions. To clarify, this is the same site where a significant accident, known as the Three Mile Island incident, took place in 1979. This unfortunate event primarily affected Unit 2, whereas Unit 1 remained unaffected. According to Constellation, Unit 2 is projected to resume operation by the year 2028.

If Constellation and Helion’s plans come to fruition, it would mean that the practical nuclear fusion operation is just a few years away from Microsoft’s ambitious goal of becoming carbon-neutral by 2030.

Microsoft’s power consumption has never been higher

What is the progress Microsoft has made towards their carbon-neutral objectives? It appears that, viewed from a limited perspective, they have some challenges to overcome.

According to a report by Michael Thomas, it was found that Microsoft used approximately 24 Terawatt-hours (TWh) of electricity in the year 2023. This amount of energy is greater than what over 100 countries consume. Google, too, reported the same level of energy consumption as Microsoft for the same time period.

In May, Microsoft released its 2025 Environmental Sustainability Report, which contained significant figures that posed challenges for Microsoft’s ambition to achieve carbon negativity by 2030.

Microsoft experienced a substantial growth, with carbon emissions increasing by 23.4%, primarily related to AI operations. Furthermore, there was an astonishing 168% rise in total energy consumption compared to the company’s energy demands in 2020, as reported by Thomas. These figures from the report don’t appear unbelievable at all.

Brad Smith, president of Microsoft, emphasized after releasing their sustainability report that achieving the 2030 objectives is more like running a marathon than a sprint. In simpler terms, he highlighted that it’s about making consistent progress over a longer period rather than rushing to complete everything quickly.

Just like OpenAI, the innovative company behind ChatGPT and heavily backed by Microsoft, faces a comparable energy predicament. In 2024, OpenAI CEO Sam Altman suggested that the escalating energy requirement of AI can only be addressed through nuclear fusion.

In August 2025, OpenAI unveiled its newest GPT-5 model, which turned out to be a powerhouse in terms of energy consumption – the highest yet. As per findings from the AI lab at the University of Rhode Island, one query with GPT-5 can consume up to 18 watt-hours of energy.

Approximately 2.5 billion daily requests keep GPT-5 running, requiring around 45 gigawatt-hours (GWh) of energy. This is equivalent to the electricity consumption of approximately 1.5 million households in the United States. To illustrate this, a modern nuclear power plant generates between 1 and 1.6 gigawatts (GW) of power per reactor per hour.

It’s a topic of debate on just how much energy is required to power generative AI, but the key point is that our existing energy resources are struggling to meet the demands.

Why use nuclear energy to power AI data centers?

It’s common for people to have misconceptions about nuclear energy. Essentially, there are two main processes: nuclear fusion and nuclear fission. In nuclear fusion, atoms combine, while in nuclear fission, they split. Regardless of the process, both produce heat. This heat causes water to turn into steam, which spins turbines. The rotation of those turbines generates power.

Given that the main by-product of nuclear energy is water vapor, it appears to be an ideal solution for businesses requiring substantial power with minimal carbon emissions. This is likely one of the reasons why a forward-thinking company like Microsoft finds it appealing.

While we may focus on the benefits of nuclear energy, it’s essential to acknowledge two significant challenges: the management of nuclear waste generated by fission, a topic of great concern due to its safe storage and disposal; and the risk of a meltdown, although infrequent, poses an immense threat in terms of both human casualties and environmental damage.

As a researcher myself, I must clarify that while nuclear fusion doesn’t generate the same amount of radioactive waste as fission reactions, it remains in the experimental phase, preventing us from reaping its full potential at this moment. As Alex de Vries, a data scientist and researcher at Vrije Universiteit Amsterdam, aptly put it earlier, considering fusion energy to meet AI demands is more like wishful thinking at present.

Read More

2025-09-09 17:41