Loudoun County in northern Virginia was once known for its vast farms and historic landmarks from the American Civil War.
The last 15 years fields and forests they were gradually freed up to create some of the enormous data centers that form the backbone of our everyday digital lives.
Just one of the newly constructed facilities will, once completed, require as much energy as approximately 30,000 U.S. homes consume.
Around the world today there are more than 7,000 data centerswith various stages of development starting from 3600 which was in 2015.
It is estimated that by 2034 the energy consumption of data centers will have reached 1,580 TWh (terawatt hours) as much as it consumes today all over India.
Energy, here and now
Artificial Intelligence means more demand for processing power, therefore more energy, but also more land to build new and increasingly larger data centers.
“That’s the thing about AI. She needs a lot of energy and she wants it now”, say experts.
A sudden increase in electricity demand from data centers is now outstripping available power supplies in many parts of the world, according to data center managers, energy suppliers and technology executives.
The dramatic increase in energy needs due to its trend Silicon Valley for development at all costs in the fields of artificial intelligence also threatens to derail the energy transition plans of entire nations and the clean energy technology programs where it has been invested trillion dollar technology.
There are some countries, such as Saudi Arabia, Ireland and Malaysia, that have the energy required to operate all these data centers that are planned to be built will exceed the available supply of renewable energy.
O Sweden Also, for example, you could see data center energy demand roughly double over this decade, and then double again by 2040.
Scary numbers
In the UK, AI is expected to absorb 500% more energy in the next decade. And in the US, data centers are predicted to use 8% of total energy by 2030, up from 3% in 2022, according to Goldman Sachs.
And these are all just estimates. As there is still a lot of uncertainty about what the current AI frenzy will look like.
For their part, technology companies are defensively quick to point out that data centers represent less than 2% of global energy consumption, even with all the current expansion.
However, a recent report from Goldman Sachs estimates that this percentage could increase to 4% by the end of the decade.
Any percentage point increase is giant, given that global electricity demand has remained almost constant for years, if not even declining in some areas.
In the U.S., electricity demand is expected to grow 40% over the next two decades, compared with an increase of just 9% over the past 20 years, according to John Ketchum, chief executive of NextEra Energy Inc., the largest wind generator of the world. and solar energy without government participation.
“It’s artificial intelligence”
Data centers are the biggest reason for this boom in demand, says Ketchum
And when asked why data centers are suddenly absorbing so much energy, his answer is blunt:
“It’s artificial intelligence,” he says, referring to energy needs which require training models, but also the inference process by which artificial intelligence extracts results from data.
“It’s 10 to 15 times the amount of electricity.”
In total, data centers use more electricity than all countries on the planet
Only 16 countries, including the US and China, consume more energy.
The largest cloud service providers, Amazon, Microsoft and Google, have announced goals to operate their data centers entirely on green energy, Amazon by next year, Google and Microsoft by 2030.
And in the background nuclear power plants
All three say they are working on technological methods to use less energy or balance demand on the grid more efficiently. This may include limiting the performance of chips and servers, arranging equipment in a demanding manner less cooling and shifting loads to different areas based on where energy is available, especially green energy.
However, some leading figures in the technology space, such as OpenAI CEO Sam Altman, say a new energy breakthrough – likely from nuclear power – will likely be needed to adapt to the new data.
Microsoft and Amazon They’re also betting on nuclear power, with Amazon recently purchasing a nuclear-powered data center in Pennsylvania and looking for more.
For now, the path forward remains unclear. Microsoft recently admitted that the whole AI issue is compromising its long-term goal of becoming carbon-free by 2030.
The quiet farms of Virginia,,,
In addition to the energy suffocation that Artificial Intelligence threatens, there is another parameter that takes us back to Virginia farms: data centers increasingly need space. And countries like Ireland already face this problem.
Virginia’s expropriated farms will not (be) alone.