Data Centers on the Brink – AI’s Growing Hunger for Power
Neural networks no longer just generate texts — they draw massive amounts of electricity. The more powerful the model, the more energy it needs. Startups, big tech giants, research labs — all are offloading computations to the cloud. Data centers are booming, but energy grids can’t keep up with the pace.
AI has become one of the main drivers behind the new surge in global energy consumption. While one artificial intelligence composes poetry, another already demands a dedicated power supply line.
Previously, all this infrastructure consumed around 1.3% of global electricity usage. Now the numbers are very different, though no one names exact figures yet. The load has shifted. It’s not just about the number of servers anymore — it’s the entire operating principle that’s changed. Cloud hubs that once helped prevent overloads are now creating them.
The potential for saving energy is nearly exhausted. Everything that could be squeezed out has already been used. There are no easy fixes left. Every new request to an AI is another draw from the shared energy grid. And those draws keep increasing.
The figures may seem alarming, but it’s too early to panic. Even with AI in the mix, data centers won’t become the top consumers. Their share in overall growth is still under 10%. Electric vehicles and household cooling are advancing faster and consuming more. For now.
But the issue isn’t about scale — it’s about concentration. On average, the situation is manageable, but in certain areas it’s already stretched to the limit. Some networks are holding up, others are starting to buckle. After all, AI isn’t the only pressure point on the global internet infrastructure. The rise of online gaming and iGaming has added even more strain. Mobile devices have become the main platform for these types of entertainment thanks to user-friendly casino apps. The most popular among players are modern mobile arcade games. We reviewed several thematic aggregator sites and found that aviatorapps.in offers apps from various online casinos featuring Aviator and other crash games. These games require a constant internet connection.
In places like Virginia, Ireland, and Amsterdam, everything is hanging by a thread. Data centers are clustering near undersea cables, where taxes are lower and skilled labor is available. As a result, the load on the same regions has reached critical levels.
In some areas, temporary construction bans are being introduced. In others, power is simply being shut off. These aren’t isolated incidents — they’re early warning signs. If data centers with gigawatt-scale capacity emerge, overload will no longer be a metaphor.
The solution sounds simple — things need to speed up. As long as the pace of building new centers outstrips the development of the networks, one has to be adjusted to match the other. The International Energy Agency advises against overcomplicating things: share forecasts, coordinate efforts, and launch more clean energy capacity.
Otherwise, everything will start to stall. Trouble spots will become permanent, and those who fall behind will end up paying the price. In the coming years, it will become clear whether AI will get its own dedicated energy source or keep struggling with power shortages.
See also: How Cloud Technologies Contribute to Environmental Sustainability