As AI companies race to secure power for their data centers, natural gas is having a moment.
Data centers are growing bigger and more power-intensive. They are being built on tighter timelines than ever before, and the current power infrastructure in the United States is struggling to keep up.
“Speed matters,” Vivian Lee, a managing director and partner at Boston Consulting Group, told Business Insider. “Bringing a site online even a year earlier can have a meaningful economic impact.”
Lee said it typically takes two to three years to build a data center, assuming the local community is on board. If they aren’t, it could take longer. At the same time, grid upgrades can take four to eight years, so AI companies are looking for faster ways to secure power, Lee said.
One of the quickest ways to do that these days is to use existing natural gas infrastructure. Gas plants can often be built or expanded faster than nuclear projects, plug into an extensive pipeline network, and provide greater energy security than renewable resources.
Meta is adding seven natural gas plants to its Hyperion data center in Louisiana. Chevron and Engine No. 1, which announced a partnership last year to build natural gas-powered plants next to US data centers, recently brought Microsoft into the fold and are now working to power a data center campus in West Texas. Google has also confirmed that it is working with Crusoe Energy to build natural gas plants to partially power its “Goodnight” data center campus in the Texas Panhandle.
“The most important metric now is speed to power — and a lot of it. That’s why gas is back in focus,” Jamie Webster, a senior director and partner at BCG, said.
Renewables take a back seat
Natural gas produces less carbon dioxide per unit of energy than coal or oil when burned, but it is still a fossil fuel and a driver of the climate crisis.
That makes Silicon Valley’s recent embrace of natural gas notable.
American tech leaders have long positioned themselves as leaders in the shift to renewables. Tech giants, including Google, Amazon, Microsoft, and Meta, signed massive wind and solar power deals as recently as last year to offset the growing electricity demand of their data centers.
Those deals, however, may have been more about money than principle.
Webster said the focus on renewables has been driven not only by sustainability goals, but also by their ability to lower energy costs over time, since they’re the only energy source whose costs tend to decline.
“Over the past decade, technologies like solar, wind, and batteries saw cost reductions of up to 90%, which fundamentally changed the equation,” he said.
The bottom line is also what’s now driving the shift back to natural gas, he said, especially as AI companies raise massive amounts of capital to build the technology’s infrastructure while still showing little in comparative revenue.
In conversations with developers and energy providers, Lee said renewables are seen as “essential, but not sufficient on their own.”
Carbon capture technology could help blunt the environmental impact of more natural gas use. The process traps carbon dioxide from power plants or industrial facilities before it reaches the atmosphere, then stores it underground or reuses it. That could allow companies to keep using natural gas while reducing the emissions that make it a major contributor to the climate crisis.
Webster, however, said that carbon capture technology — sometimes known as CCUS, or “Carbon Capture, Utilization, and Storage” — is still in the early stages of scaling.
The world has now entered a “structural supercycle,” driven in part by data centers, electrification, and cooling demand, Webster said.
“That growth is putting pressure on supply, and in many cases, gas is one of the fastest ways to meet it.”
Read the original article on Business Insider
The post The AI industry is turning to an old energy favorite to bring its data centers online fast appeared first on Business Insider.
