Last week, Microsoft announced that it would spend approximately $80 billion during this fiscal year to build data centers for its booming artificial intelligence business.
That gargantuan sum is a testament to the opportunity that Microsoft and other tech giants see in A.I.
It also has the makings of a climate conundrum.
In order to power all of those data centers, Microsoft and other tech companies building similar projects are going to need huge amounts of electricity. That means ever greater strains on a power grid that is still, to a significant extent, fueled by natural gas, and to a lesser extent, coal.
And, as the A.I. boom continues, we’re starting to learn that it will mean more planet-warming emissions, at least in the short term.
In the next three years alone, data centers are expected to as much as triple their energy use, according to a new report supported by the U.S. Department of Energy. Under that forecast, data centers could account for as much as 12 percent of the nation’s electricity consumption by 2028. McKinsey & Company, the international consulting firm, expects global demand for data centers to grow at roughly 20 percent a year through the end of the decade.
Climate contradictions
There’s rich irony in the fact that it is Microsoft and other tech companies that are responsible for a surge in energy demand, and thus an uptick in planet warming emissions.
For the past decade, these same companies have been leading corporate America’s efforts to combat global warming. There was Microsoft’s promise in 2020 to remove all of its historic emissions from the atmosphere. And Amazon’s climate pledge. And Google’s net-zero goals.
Now those same companies are watching their climate targets recede thanks to their big bets on A.I.
Microsoft and Google have both reported sharp jumps in their emissions in recent years because of the soaring energy needs of data centers.
While that may seem like the makings of a giant contradiction, it could turn out to be a good thing, in the long term, for the future of clean power.
A.I. evangelists at these companies say their technology is poised to improve energy efficiency around the world and deliver breakthroughs in the production of clean power. If this comes to pass, today’s spike in emissions will be just a short-term disruption to their climate goals.
And while the tech giants are hungry for power now, they remain big supporters of renewable power and big buyers of carbon credits.
Microsoft did not say how much of its 2025 investments would be offset by renewable energy investments or credits. But Bobby Hollis, the company’s vice president of energy said in a statement that Microsoft “embraces a multi-technology approach, as outlined by the International Energy Agency, with a broad range of carbon-free technologies.”
Nili Gilbert, the vice chairwoman of Carbon Direct, a company that helps corporations lower their emissions, said it was “encouraging” that “the companies seeing growth in A.I. data centers are also the most highly committed and the biggest spenders on their climate targets.”
“They’re very focused on renewable energy,” she added.
A nuclear renaissance?
The scramble to meet the energy demands of data centers is leading to a flurry of investments in nuclear energy.
Last year, Microsoft said it would pay a company to revive the shuttered Three Mile Island nuclear power plant in Pennsylvania. Google and Amazon said they were betting on newer, smaller reactors that could become widely available later in the decade.
And Sam Altman, the chief executive of OpenAI, has responded to queries about the energy demands of A.I. by expressing his faith that nuclear fusion, which has only worked sporadically in the lab, is on the cusp of going mainstream.
“Fusion’s gonna work,” Altman told Bloomberg Businessweek in a recent interview.
(Altman is an investor in a fusion startup, Helion, one of several companies that say they are on the cusp of an energy revolution, as my colleague Raymond Zhong recently reported.)
But many of the nuclear investments being made today, even the conventional ones, will take several years to pay off.
“The challenge is, this is a decade away, and they need power in 12 or 18 months,” said Jason Bordoff, the founding director of the Center on Global Energy Policy at Columbia University. “And there are few options to do that in the very short term other than natural gas.”
Natural gas
Gas is the largest source of power for the U.S. grid, accounting for about 42 percent of electricity generation. And given the number of gas-fired power plants already out there, it’s also the fuel source that is easiest to scale up quickly to meet the additional demand coming from data centers.
But burning natural gas releases potent planet-warming emissions. While it emits less carbon dioxide than coal and oil, burning gas releases lots of methane, which traps heat in the atmosphere.
A new generation of technology, carbon capture and storage (C.C.S.) has the potential to make natural gas less polluting. But carbon capture technology is expensive, and its widespread deployment is a long way off.
Still, with data centers hungry for more energy and nuclear and renewables unable to meet the demand, even some experts see an important role for natural gas in the short term. “Natural gas plus C.C.S. is one of the quickest ways that we can bring low-carbon base load power online,” Gilbert said.
The promise of A.I.
In the long term, could artificial intelligence help solve climate change? That’s the hope, and it’s a vision shared not just by the executives who stand to profit from the A.I. boom.
Whether it’s making traffic patterns and airline routes more efficient, optimizing the way electricity flows through the power grid or identifying new technological breakthroughs, A.I. proponents say that the technology has the potential to help save more power than it consumes.
“On the one hand, A.I. is going to take energy, and on the other hand, A.I. can help improve how we deploy clean energy,” Bordoff said. “I am optimistic that that’s going to net out in a positive direction from an energy transition standpoint.”
That’s an rosy view of how the A.I. revolution will unfold.
But until nuclear takes off, or fusion comes through, or A.I. unleashes a new wave of cheap renewables, artificial intelligence will exact a steep toll on the climate.
The Panama Canal’s newest voyagers: fishy intruders from two oceans
The Panama Canal has for more than a century connected far-flung peoples and economies, making it an essential artery for global trade.
But of late the canal has been linking something else, too: the immense ecosystems of the Atlantic and the Pacific.
In less than a decade, fish from both oceans — snooks, jacks, snappers and more — have almost entirely displaced the freshwater species that were in the canal system before, scientists with the Smithsonian Tropical Research Institute in Panama have found. Fishermen around Lake Gatún who rely on those species, chiefly peacock bass and tilapia, say their catches are growing scarce.
Researchers now worry that more fish could start making their way through from one ocean to the other. — Raymond Zhong
More climate news:
-
Climate change is “wreaking havoc” on the planet’s water cycle, with ferocious floods and crippling droughts affecting billions of people, according to a new report highlighted by The Guardian.
-
Heatmap News reports on the money pouring into sustainable aviation fuel, and why aviation industry has been so hard to decarbonize.
-
Global warming is moving faster than the world’s best climate models, The Atlantic reports.
Thanks for being a subscriber.
Read past editions of the newsletter here.
If you’re enjoying what you’re reading, please consider recommending it to others. They can sign up here. Browse all of our subscriber-only newsletters here. And follow The New York Times on Instagram, Threads, Facebook and TikTok at @nytimes.
Reach us at [email protected]. We read every message, and reply to many!
The post What Will Power the A.I. Revolution? appeared first on New York Times.
