As artificial intelligence (AI) becomes the focus of the technology world, few people realize that the smart tools we use every day.
From chatbot to video creator, all are powered by energy exploited from da and fossil fuels.
Behind that modern shell are flat-lined areas, depleted water resources and affected communities.
According to the Wall Street Journal, startup Poolside is building a Horizon data center complex of more than 500 units in West Texas, nearly 500 km from Dallas.
The facility will produce its own electricity by burning natural gas exploited from the Permian basin, the most productive oil and gas field in the United States.
When completed, Horizon could generate two electricity gigawatt, equivalent to the Hoover Dam, but at a significant environmental cost.
AI giants such as OpenAI, Meta and Elon Musk's xAI are all pursuing a similar strategy.
OpenAI's Stargate data center in Abilene, Texas, needs up to 900 megawatt of electricity for eight buildings and is supported by a newly built gas power plant.
Meta is investing10 billion to build a giant data center in Louisiana, consuming 2 gigawatt of electricity, all supplied from three new natural gas plants.
These projects bring economic benefits, but also make local people feel affected.
Ms. Arlene Mendler, who lives opposite the Stargate project, said that the peaceful land where she has been attached to for more than 30 years has disappeared: "We used to come here to find quiet, but now there is only the sound of the wind and the bright lights all night long".
In addition to the noise, Texans are worried about water sources, as the data center consumes millions of gallons of cooling systems in areas with frequent droughts.
Researchers warn that the rapid growth of AI data centers is driving the dependence on fossil fuels, increasing greenhouse gas emissions.
Although companies like Meta are committed to clean energy, most of their facilities still rely on natural gas because it is more practical and cheaper.
Mr. Chris Lehane, Vice President of OpenAI, said that the US needs to produce about one gigawatt of new energy per week to meet computing needs.
However, Duke University experts point out that power companies currently only use about 53% of available capacity, and flexible coordination can completely meet AI needs without opening more gas plants.
Amid the wave of hundreds of billions of dollars invested in data centers, the big question remains unanswered: Is humanity trading too much to nurture artificial intelligence?
As rural areas are flat, clean water is depleted and gas continues to be burned, the real benefits of the AI era may not be fairly distributed, and it is those who do not require AI who are the first to pay the price.
