As artificial intelligence develops rapidly, the energy consumption issue of AI applications has become an increasingly prominent concern in the global academic and industrial circles.

Recently, more and more large tech companies are turning to nuclear power to meet the increasing power demands of their datacenters, while $NVIDIA (NVDA.US)$ CEO Huang Renxun also mentioned that nuclear power may be a feasible choice to meet the energy needs of AI.

Nuclear power is a feasible choice for meeting the energy needs of AI

On Friday, US East Coast time, Huang Renxun stated in an interview that an increasing number of datacenters require renewable energy, and nuclear power is a great option.

Huang Renxun said: 'Nuclear power, as a form of energy, a sustainable energy, is excellent. But it won't be the only source of energy. We will need all sources of energy and balance the availability, cost, and long-term sustainability of energy.'

He hinted that Nvidia may be interested in nuclear power, which is part of its broader strategy to diversify energy sources for datacenters.

According to the International Energy Agency (IEA) data, by the end of 2026, global data centers' electricity consumption is expected to increase to between 650TWh and 1050TWh, nearly doubling from 460TWh in 2022. Therefore, it is crucial in the long term to find cleaner and greener alternative energy sources than fossil fuels.

In some parts of the world, electricity generation is already insufficient to support the construction of new data centers, and this issue is becoming increasingly severe. Huang Renxun mentioned that energy shortages have affected technology companies' considerations for the location of new 'AI factories.' Some data centers are unable to operate at full capacity, while others can only be built far from population centers.

The benefits of AI are expected to outweigh its massive energy consumption.

Huang Renxun also believes that the benefits of artificial intelligence to the grid and society will ultimately offset its huge energy consumption.

Currently, Nvidia's latest generation of chips consumes over 1 kilowatt per chip. Huang Renxun stated that although Nvidia's new products consume more power than the previous generation, they are more efficient, allowing them to train and run AI software faster, which is beneficial in the long run.

In addition, Huang Renxun also mentioned that he is making every effort to serve Chinese customers and comply with the US government's restrictions on Nvidia exports to China.

He said, "The first thing we have to do is to comply with any policies and regulations. At the same time, in the markets we serve, we must compete to the best of our abilities. There are many customers who rely on us, and we will do our utmost to support them."

Editor/Rocky