Is Google’s data center location strategy at odds with its emissions goals?

Not all renewable energy projects are equal in their emissions impact. A solar project built in eastern Kentucky, for example, where the grid is dirtier, has a bigger impact on avoided emissions than one built in California.

And the inverse is true for data centers. Given their immense energy demands, a data center located in a place where the grid draws on mostly renewable power will emit far fewer emissions than one connected to a dirty grid. 

Locational data can be used to optimize where a renewables project can prompt the most dramatic emissions reductions — and where a new data center or manufacturing facility can have the lowest emissions toll.

• The top line: Big Tech is starting to use the available data to optimize its computing and data center siting, said WattTime Executive Director and Co-Founder Gavin McCormick. But Google is an outlier, seeming to make siting decisions primarily based on local energy access, rather than energy demand or emissions.
• The current take: In reviewing the data, McCormick said that WattTime began “noticing a pattern that data centers and renewable energy projects not owned by Google are systematically getting built in smarter places than they used to be.” And those impacts are even clearer when comparing renewable-fossil fuel energy mixes around the country and the locations of Google’s data centers, he added.

As the need for data centers grows amid the artificial intelligence boom, terawatts of new load growth are expected in the coming decades. This new wave of demand — and associated emissions — has left large tech companies scrambling for new locations to build warehouse-scale computers. 

There are essentially two competing methods for how companies site their new computing projects, McCormick said: Google versus everyone else. Google, more than its peers, makes decisions based on local energy access, he said. 

“I think it comes from a good place, but there’s so much evidence now that [building locally] isn’t necessarily better for climate change, as the emissions impacts of a project might be significantly different elsewhere,” McCormick said, adding that Google’s strategy likely isn’t saving them any money. “They just feel like local energy is a good thing.”

On the contrary, he says, other tech companies like Amazon, Meta, and Microsoft are starting to rely on data to optimize their data center siting for prime emissions reductions. McCormick has spoken to all of them, and this is the approach that he recommends.

Google has a goal of operating on 100% carbon-free energy by 2030. And the company objects to McCormick’s characterization. 

“We have a very rigorous process that we use to select sites, which includes…the availability of (or potential to bring) carbon-free energy,” a spokesperson told Latitude Media, adding that as the company pursues its 24/7 carbon-free energy goals, “our focus is on procuring clean energy to support decarbonizing the grids where we operate.”

This divide has emerged across 24/7 clean energy procurement more broadly. WattTime is an “allied organization” of the Emissions First Partnership, created by companies including Meta and Amazon to prioritize emissions reductions over any specific time-matching structure. Google and Microsoft are instead a part of the Granular Certificate Trading Alliance, which champions building a marketplace for hourly certificates. 

WattTime and the problem of ignored data

The lack of locational data is also a problem in the renewables sector, McCormick said. Projects aren’t being built in the most effective locations for emissions savings.

“The place that people build the most renewable energy is by far the least impactful place in the entire country to build renewable energy,” McCormick said, adding that northern California solar avoids the least amount of carbon emissions nationwide. “I’m telling you definitively that developers are not considering that locational data in their decisions.”

In partnership with the company Clearloop, WattTime recently published a white paper that added to the mounting pile of locational data for developers. The paper outlined the use of marginal emissions data to measure the impact of a project built in a renewables-rich area versus one in an area that is heavily reliant on fossil fuels, and to determine the most “climate-friendly” location.

(Clearloop, which was acquired by independent power producer Silicon Ranch, measures other companies’ carbon footprints and offsets them by investing in renewables projects in places where locational data suggests would have the highest impact.)

WattTime’s marginal emissions data is already available nationwide, and the nonprofit expects to roll out its global marginal emissions data within weeks. But the real question is whether more data will make a difference.

“We are a nonprofit because no one is asking for this data,” McCormick said, adding that there’s little to no market for companies who want to drive more impact. McCormick explained that back in WattTime’s early days, he and other graduate students immersed themselves in the developer landscape between interviews, internships, and anything they could do to see “how all the decisions really were made.”

“We could not find a single renewable energy developer who was actually factoring in the emissions impact in deciding where to build,” he said, adding that he considers that “absolutely bonkers.”

McCormick’s theory? Developers have a misperception about their impact.

“There’s a lot of confusion around the idea that ‘surely if we’re doing the more profitable projects, surely that’s because the market is telling us it’s better and that it’s impactful,’” he said. “The idea that you actually need to look at the emissions is very new and different.”