The processing capacity needed to support applications such as ChatGPT and generative AI systems has put conventional electrical grids in check. This has increased interest in stable, clean and high-performance energy sources.
In this context, Nvidia, a giant in the development of hardware for AI, is increasingly betting on nuclear energy as the ideal solution for its energy needs.
This synergy between Nvidia’s technology and the potential of nuclear energy offers a viable and attractive alternative to fossil and intermittent resources, such as solar and wind energy.
Nvidia’s vision for nuclear energy
Nvidia CEO Jensen Huang has publicly stated his support for nuclear power as a key resource to meet the power needs of AI data centers.
According to Huang, nuclear power is “wonderful” in terms of sustainability and efficiency for large data centers, and although Nvidia does not have its own cloud infrastructure, the company faces a common challenge in the industry: the constant and affordable supply of electricity.
AI technology and energy demand
With Nvidia as the market leader in GPUs (graphics processing units), especially those used in large AI data centers, the power demands are colossal.
These facilities must remain active 24/7 and without interruptions, and with intensive use of electricity that cannot always be sustained by intermittent renewable sources.
Nuclear power, in contrast, offers continuous, large-scale supply, making it especially attractive to meet the voracious energy demand of generative AI. But generative AI may also have something to offer the nuclear energy field.
The case of Diablo Canyon and “Nuclear AI”
One of the projects that exemplifies this synergy is the recent collaboration between Pacific Gas & Electric (PG&E) and Nvidia-backed startup Atomic Canyon to deploy AI tools at the Diablo Canyon nuclear plant in California.
Through the Neutron Enterprise system, a specific generative AI for the nuclear sector, it is expected to optimize the management of documentation, procedures and maintenance of the plant.
Neutron Enterprise, built on the Nvidia platform, can reduce search times across millions of technical documents, enabling greater operational efficiency and complying with federal regulations.
This project marks the first deployment of generative AI in a nuclear facility in the United States and reinforces the relevance of AI as an improvement tool in the management of nuclear plants.
Nvidia in Japan: Datacenters powered by nuclear energy
Another front where Nvidia explores this synergy is Japan. The company Ubitus, backed by Nvidia, plans to build a new data center in the Asian country, which would be located near nuclear plants.
This center would be dedicated to generative AI services and requires a constant, large-scale energy supply that renewable energies cannot offer.
This initiative, along with similar projects from other large technology companies, shows how AI data centers seek to increasingly align themselves with nuclear energy as a viable, sustainable and scalable source.
The AI-driven nuclear energy resurgence
Rising demand for electricity in AI data centers is contributing to renewed interest in nuclear energy.
From the restart of plants like Three Mile Island in Pennsylvania to power Microsoft’s AI systems to developments in small modular nuclear reactors (SMRs), nuclear energy is experiencing a new boost driven largely by the needs of AI.
These modular reactors, which offer advantages such as smaller size, portability and reduced costs, are in full development and have attracted the attention of investors and technology companies.
Nvidia, along with other giants such as Amazon, Microsoft and Google, is evaluating its potential to meet the growing demands of its data centers.
Will Nvidia invest in the nuclear sector?
Given that Nvidia has already invested in other technology startups, it is reasonable to speculate that it could consider investments in nuclear technology companies, especially in the SMR sector.
Huang, aware of the importance of nuclear power in sustaining AI infrastructure, appears to lean in favor of these partnerships. After several years of prosperity, what is clear is that they do not lack economic muscle.
SMR reactors not only enable stable supply, but are scalable and can be tailored to the specific demand of data centers.
An “old” technology that is back in fashion
Despite its benefits, nuclear energy remains a controversial topic due to concerns about safety and waste management.
In countries like Germany, where nuclear power plants have been closed, and in Spain, which is in the process of doing so, these concerns are particularly relevant.
The restart of plants like Diablo Canyon in the United States, along with renewed interest in small nuclear reactors (SMRs), raises questions about the safety and long-term sustainability of the synergy between AI and nuclear energy.
Nvidia’s recent projects, such as Ubitus in Japan, reflect the delicate balance between harnessing the benefits of nuclear energy and addressing concerns about its environmental impact.
This post is also available in: