Nvidia CEO Jensen Huang appeared on Joe Rogan’s podcast this week and made a clear statement about AI’s future. He said energy is now the main limit for AI growth, not computer chips. Huang explained that the power grid cannot keep up with rising demand from AI data centers.
Huang said large tech companies may need to run small nuclear reactors near their data centers. He expects this to happen within six to seven years. These compact plants would provide steady power directly to the facilities that need it most.
The Nvidia CEO described modern AI data centers as “gigawatt factories” that need huge amounts of electricity. He said these concentrated power needs cannot easily connect to the existing public grid. Adding them could create stability problems for the wider power network.
Huang suggested that on-site nuclear plants could help solve this problem. The reactors would be in the hundreds of megawatts range. When the data centers don’t need all the power, the extra electricity could flow back to nearby towns.
Several large tech firms have already started making deals for nuclear power. In late 2024, Alphabet announced it would buy 500 megawatts from Kairos Power. The company plans to have the first unit running by 2030.
In mid-2025, Kairos Power and the Tennessee Valley Authority made another deal. They agreed to build a small plant in Oak Ridge that will generate 50 megawatts. This power will go to sites that support Google’s data operations.
These deals show that tech companies are taking the energy problem seriously. They are looking for long-term solutions that can provide reliable power at large scale. Nuclear energy offers this combination in ways that other power sources may not.
Data from major financial firms shows why tech companies are worried about power. Goldman Sachs predicts data center energy use will jump 175% by 2030 compared to 2023 levels. This growth reflects the rapid expansion of AI tools and cloud computing services.
The International Energy Agency has similar projections for global energy demand. The agency expects data centers worldwide to use 945 terawatt hours by 2030. This is up from 415 terawatt hours in 2024, more than doubling in just six years.
Both organizations identify grid capacity as a key challenge for the coming years. The existing power infrastructure was not built to handle such concentrated demand. Traditional grids spread power across many different users and locations.
Nuclear stocks dropped in November as investors questioned how these companies would fund their projects. Small nuclear reactor developers like NuScale Power, Nano Nuclear Energy, and Oklo saw their share prices fall. These companies currently have little revenue but face expensive, multi-year development timelines.
Huang’s comments may help these companies regain investor confidence. He connected small nuclear reactors directly to AI growth, one of the hottest sectors in technology. His six to seven year timeline gives investors a clear timeframe to watch for progress.
The statement from Nvidia’s CEO validates the business case for small modular reactors. It confirms that demand exists and comes from companies with deep pockets. Tech giants need reliable power solutions and appear willing to invest in nuclear technology to get them.
NVIDIA Corporation, NVDA
Nvidia’s stock closed at $182.41 on Friday, down slightly for the day. Kairos Power and the Tennessee Valley Authority completed their Oak Ridge deal earlier this year.
The post Why Nvidia’s Jensen Huang Thinks AI’s Future Depends on Nuclear Power appeared first on CoinCentral.


