In recent months theArtificial intelligence has experienced a real boom, with heavy investments by the technological giants and the construction of dedicated data centers all over the world.
The exponential growth of this technology, however, has also significantly increased the energy consumption, with facilities experiencing power issues. In this sense, we are now talking about real “bottlenecks” that are slowing down the growth of AI.
This is the scenario described by Financial Timeseffectively confirming what was predicted a few days ago by Elon Musk. The entrepreneur, in fact, in a recent interview, had stated how after the problems related to the lack of chip of 2023, the next brake on the development of these technologies would have been linked to energy supply.
An alarm that is echoed when said a few months earlier by Andy Jassythe CEO of Amazon.
Data centers and energy crisis: Microsoft would even evaluate small nuclear power plants
In a historical moment where we talk about sustainability, the consumption linked to the maintenance of AI is a real punch in the stomach. According to the IEA, the International Energy Agency, in 2026 electricity consumption in data centers around the world is expected to increase to 1.000 TWh. To quantify this value, just think that it is the total electricity consumption of all of Japan.
The simple operation of data centers, however, is not the only problem related to AI. In fact, these structures cannot be built everywhere: access is needed water sources to be able to cool devices. The places suitable for this purpose, especially in the American context, have already been occupied.
At present, in fact, lack of energy appears to be the only major obstacle facing the development of AI. To understand how strong the energy “thirst” is, just think that MicrosoftIn the past year, it began exploring the idea of deploying small nuclear reactors to cool its data centers.