Filenews 20 February 2026
The explosive growth of AI is transforming electricity from operational costs into a strategic competitiveness factor, prompting tech giants like Microsoft and Amazon to re-turn to nuclear power to support their infrastructure.
The oversized data centers required to train and operate advanced AI models consume 300 to 500 megawatts each, a power comparable to that of a medium-sized city. With dozens of such facilities under development, ensuring a steady, 24-hour electricity supply becomes crucial.
Energy as a strategic asset
AI systems operate uninterruptedly, rendering intermittent renewables insufficient without supplementary stationary generation or large-scale storage. Thus, electricity ceases to be a secondary expense and emerges as a structural constraint of technological development.
For years, tech companies relied on renewable energy credits to reduce their carbon footprint. However, the increasing demand is making this practice less viable, reinforcing the need for reliable sources of base.
Microsoft's Nuclear Shift
Microsoft is involved in the restart of the former Three Mile Island Unit 1 reactor, renamed the Crane Clean Energy Center, in partnership with Constellation Energy. The project is financed with a $1 billion loan from the US Department of Energy and aims to be commercially operational in 2027, offering long-term energy security without carbon emissions.
At the same time, the company has entered into a power purchase agreement with Helion Energy, investing in the future prospect of nuclear fusion.
Amazon's Control Strategy
Amazon opts for a more vertical approach. Through the acquisition of the Cumulus Data Center by Talen Energy, it secured direct access to the nuclear power of the Susquehanna nuclear plant, reducing reliance on congested transmission networks.
At the same time, it is investing in small modular reactors, through partnerships with Energy Northwest and X-energy, targeting almost one gigawatt capacity.
Why nuclear power is coming back
Nuclear energy offers features that align with the needs of artificial intelligence:
- Capacity factor over 90%
- Continuous production for constant loads
- Minimum direct CO₂ emissions
- Decades of service life
The demands of data centers are making nuclear power an increasingly attractive option for large-scale infrastructure.
Implications for Investments and Markets
The convergence of artificial intelligence and energy elevates the role of nuclear producers, such as Vistra, while also strengthening the uranium supply chain, with companies such as Cameco at the center of investment interest.
The True Limit of the AI Boom
The development of artificial intelligence brings back to the fore the natural basis of the digital economy: energy density and reliability. Microsoft and Amazon aren't abandoning renewables, but they recognize that without solid base power, the next phase of tech competition can't move forward.
