Power Hungry

Last July, a fleet of flatbed trucks—sent by the artificial intelligence (A.I.) company xAI—rolled into southwest Memphis, Tennessee. The trucks carried about 20 mobile power plants, fueled by natural gas, to meet the electricity demand of what xAI calls the world’s largest supercomputer, housed in an old manufacturing plant on more than 550 acres there. The computer’s electricity needs rival those of 100,000 homes and could continue to grow.

But the project’s Memphis neighbors aren’t happy about it. Those living nearby say it all happened with no warning, no public review, and no opportunity to understand how the community would benefit. The heavily industrial area is already home to an oil refinery, a steel mill, and chemical plants. Residents contend that the xAI project has made pollution worse in an area already enveloped in smog.

“We’re getting more and more days a year where it is unhealthy for us to go outside,” says KeShaun Pearson, president of Memphis Community Against Pollution and a lifelong resident of the area near the xAI site.

Last July, the artificial intelligence (A.I.) company xAI sent a fleet of flatbed trucks into southwest Memphis, Tennessee. The trucks carried about 20 mobile power plants that are fueled by natural gas. They were sent to meet the electricity demand of what xAI calls the world’s largest supercomputer. It is housed in an old manufacturing plant on more than 550 acres of land. The computer’s electricity needs rival those of 100,000 homes and could continue to grow.

But the project’s Memphis neighbors aren’t happy about it. Those living nearby say it all happened with no warning. There was no public review and no opportunity to understand how the community would benefit. The heavily industrial area is already home to an oil refinery, a steel mill, and chemical plants. Residents say that the xAI project has made pollution worse in an area already enveloped in smog.

“We’re getting more and more days a year where it is unhealthy for us to go outside,” says KeShaun Pearson, president of Memphis Community Against Pollution and a lifelong resident of the area near the xAI site.

xAI, owned by billionaire entrepreneur Elon Musk, is a competitor to ChatGPT, the A.I. program that took the world by storm after its release by OpenAI in 2022. ChatGPT can write poetry, fix software code, and produce detailed images, ushering in an era when A.I. is poised to revolutionize entire industries. Although some consider A.I. a boon to human productivity, others worry its powerful capabilities will lead to more widespread misinformation or even allow it to become autonomous and eventually resist human control. For now, though, as technology companies race to bring more data centers and A.I. facilities online, the industry is leading a surge in electricity demand that’s expected to continue for decades.

When you type a prompt into one of these programs, it gets sent to a data center like the one that houses xAI’s supercomputer in Memphis. There it moves through a server that performs billions of calculations in under a second to determine each word of its response.

“Just to generate one single word . . . that requires a huge amount of energy,” says Shaolei Ren, a professor at the University of California, Riverside, who researches A.I. and sustainability. As Ren points out, most of the power in the U.S. is generated from fossil fuels, such as natural gas, coal, and oil. Using electricity from these power grids can “generate a lot of carbon emissions or greenhouse gases, which can potentially warm up the planet in the long run.”

Some data centers, such as the one housing the xAI supercomputer, have installed on-site gas power plants to make up for shortfalls in the electricity grid. It may be cleaner than existing power, but it adds to the industry’s substantial carbon footprint, experts say.

Servers also get very hot as they work and require a lot of water to keep them cool enough to function. Research from Ren’s team published last year showed that using ChatGPT to write a 100-word email requires the equivalent of more than one bottle of water.

The A.I. revolution has disrupted many tech companies’ pledges to rapidly reduce their planet-warming emissions. Microsoft said its emissions had soared 30 percent since 2020 because of its expansion of data centers. Google’s emissions have risen nearly 50 percent over the past five years because of A.I. While such companies are working to offset their energy consumption through investments in solar, wind, and nuclear power, experts say the energy demand of A.I. data centers is growing faster.

xAI is owned by billionaire entrepreneur Elon Musk. It is a competitor to ChatGPT, the A.I. program that took the world by storm after its release by OpenAI in 2022. ChatGPT can write poetry, fix software code, and produce detailed images. A.I. is now poised to revolutionize entire industries. Although some consider A.I. a benefit to human productivity, others worry its powerful capabilities will lead to more widespread misinformation. There is also a fear that it could become autonomous and eventually resist human control. Technology companies are in a race to bring more data centers and A.I. facilities online. The industry is leading a surge in electricity demand that’s expected to continue for decades.

When you type a prompt into one of these programs, it gets sent to a data center like the one that houses xAI’s supercomputer in Memphis. There it moves through a server that performs billions of calculations in under a second to determine each word of its response.

“Just to generate one single word . . . that requires a huge amount of energy,” says Shaolei Ren, a professor at the University of California, Riverside, who researches A.I. and sustainability. As Ren points out, most of the power in the U.S. is generated from fossil fuels, such as natural gas, coal, and oil. Using electricity from these power grids can “generate a lot of carbon emissions or greenhouse gases, which can potentially warm up the planet in the long run.”

Some data centers, such as the one housing the xAI supercomputer, have installed on-site gas power plants to make up for shortfalls in the electricity grid. While it may be cleaner than existing power, it adds to the industry’s substantial carbon footprint, experts say.

Servers also get very hot as they work. They require a lot of water to keep them cool enough to function. Research from Ren’s team published last year showed that using ChatGPT to write a 100-word email requires the equivalent of more than one bottle of water.

The rapid growth of the A.I. industry has disrupted many tech companies’ pledges to rapidly reduce their planet-warming emissions. Microsoft said its emissions had soared 30 percent since 2020 because of its expansion of data centers. Google’s emissions have risen nearly 50 percent over the past five years because of A.I. Companies are working to offset their energy consumption by investing in solar, wind, and nuclear power. Experts say the energy demand of A.I. data centers is growing faster than those investments.

With A.I. technology continuing to develop, however, experts are optimistic that its energy efficiency will increase as its hardware and algorithms improve.

“Car emissions were a lot worse 100 years ago, and now they’re more efficient,” Ren says. “The same thing will happen with A.I. sooner or later.”

For instance, in January the Chinese company DeepSeek announced that it had created a powerful A.I. model that uses significantly less energy than leading U.S. models like xAI. But experts are unsure whether it will be adopted widely enough in the U.S. to significantly change energy demands from the nation’s A.I. data centers.

And many argue that A.I.’s ability to process huge amounts of data can actually help in the fight against climate change. For instance, A.I. programs are already detecting ocean litter and helping the mining and gas industries track and reduce their emissions.

But, experts say, as long as A.I. is dependent on electricity generated from fossil fuels, it’s unlikely the technology’s environmental benefits will outweigh its costs anytime soon. And as technology companies seek to rapidly expand their computing power, the large, cheap pieces of land they need are often in economically disadvantaged communities, like xAI’s Memphis site.

With A.I. technology continuing to develop better hardware and algorithms, experts hope that its energy efficiency will increase.

“Car emissions were a lot worse 100 years ago, and now they’re more efficient,” Ren says. “The same thing will happen with A.I. sooner or later.”

For instance, in January the Chinese company DeepSeek announced that it had created a powerful A.I. model that uses significantly less energy than leading U.S. models like xAI. But experts don’t know if it will be adopted widely enough in the U.S. to change the energy demands from the nation’s A.I. data centers.

There is an argument that A.I.’s ability to process huge amounts of data can actually help in the fight against climate change. A.I. programs are already detecting ocean litter and helping the mining and gas industries track and reduce their emissions.

But, experts say, as long as A.I. is dependent on electricity generated from fossil fuels, it’s unlikely the technology’s environmental benefits will outweigh its costs anytime soon. The need for technology companies to rapidly expand their computing power means they need the large, cheap pieces of land. These areas are often in economically disadvantaged communities, like xAI’s Memphis site.

Supporters of xAI’s presence in Memphis point to other major tech companies now interested in establishing operations there, like Dell and Nvidia.

“An investment like this attracts a lot of new talent,” Greater Memphis Chamber CEO Ted Townsend said at an event in December.

Opponents of the data center are still reeling from what they say was a lack of community involvement in the project’s quick approval last spring.

“There is no trust. There is no real communication,” says Pearson of Memphis Community Against Pollution. The group continues to raise awareness about what it says are the data center’s harmful effects.

“These are communities that already have a lot of pollution in the air and water,” says Maggie Shober, research director for the Southern Alliance for Clean Energy, a nonprofit organization that promotes clean energy. “It’s just one additional burden.”

Supporters of xAI’s presence in Memphis point out that other major tech companies, like Dell and Nvidia, are now interested in opening operations in the area.

“An investment like this attracts a lot of new talent,” Greater Memphis Chamber CEO Ted Townsend said at an event in December.

Opponents of the data center are still upset from what they say was a lack of community involvement in the project’s quick approval last spring.

“There is no trust. There is no real communication,” says Pearson of Memphis Community Against Pollution. The group continues to raise awareness about what it says are the data center’s harmful effects.

“These are communities that already have a lot of pollution in the air and water,” says Maggie Shober, research director for the Southern Alliance for Clean Energy, a nonprofit organization that promotes clean energy. “It’s just one additional burden.”