Big Tech Says Data Centers Must Change To Survive Local Opposition

As data center projects face fierce opposition across the U.S., leaders at Microsoft and Oracle say the industry can no longer afford to be dismissive of local concerns.

Instead, developers and hyperscalers need to be proactive and change how data centers are designed and built to be more palatable to the growing number of communities into which the industry is expanding.  

With the artificial intelligence building boom pushing large-scale data center development beyond the industry’s traditional footprint, local opposition is emerging alongside power availability as one of the sector’s biggest constraints.

Fears that data centers will industrialize residential and rural communities have become a common fear among project opponents, who warn that noise from cooling and power systems, generator emissions, demands on natural resources and the conversion of open land into massive buildings will erode their quality of life. 

Within the data center industry, the growing backlash is often framed as a messaging problem rather than a design or planning challenge.

At conferences and in trade publications, industry leaders frequently portray opposition as stemming from misinformation about issues such as water consumption or from NIMBYs’ impulsive resistance to change.

If the industry did a better job of correcting misconceptions and explaining data centers’ economic benefits, the thinking goes, communities would embrace these projects. 

But this line of thinking is a mistake, one the industry continues believing at its own peril, Microsoft and Oracle executives said at Bisnow’s Data Center Investment Conference & Expo in Ohio this month. 

While some opponents of data center projects exaggerate their environmental impacts and some of the backlash may reflect broader anxiety about AI rather than a specific proposed project, the industry’s critics aren’t wrong to voice concerns, said Tyler Riegle, senior program manager for land development at Microsoft.

Cooling systems and generators often make noise that can be disruptive. AI campuses bring large-scale industrial buildings and transmission infrastructure to areas where nothing of that kind existed previously. They use massive amounts of power and, in certain cases, water. 

Rather than instinctively dismissing opponents as misinformed NIMBYs, the industry needs to invest in redesigning data centers and changing development practices to proactively address local concerns, Riegle said. 

These are solvable problems, but the industry as a whole has been behind the eight-ball when it comes to making data centers a better fit for the residential and rural communities where AI development is increasingly taking place, he said. 

“The scale and demand of these facilities just grew at a rate where we weren't able to keep up with a lot of those concerns,” Riegle said. “The industry is catching up to what it means to build a 300-acre, 500-megawatt data center in a rural community.”

Data center developers should get ahead of local concerns with infrastructure and commitments to add to communities and ease the burden on local resources, he said.

Hyperscalers and data center firms should never have been caught off guard by the wave of resistance that emerged as the industry expanded into new territory, said Oracle’s Craig Deering, an architect and the firm’s senior data center construction manager. 

From an architectural perspective, the backlash is entirely predictable, Deering said. It is a well-established design principle that introducing large, “urban” buildings into residential or rural areas often triggers concerns about noise, emissions, visual impact and resource consumption. 

Deering said the industry didn’t adapt how data centers were designed and integrated into their surroundings as that expansion occurred. The result was a wave of negative publicity and cratering public perception that will be far more difficult to reverse than it would have been to prevent. 

“As an architect, I believe the design community has failed the industry on this point,” Deering said. “Native AI data centers, because of their intensity, are urban projects, so the second you take an urban project to an exurban location, a rural location or an agrarian location, you have a scale mismatch, and the design community should have advised all these owner-operators of that challenge.”

Many of the community complaints about data centers are surmountable, Deering said. In most cases, solutions already exist. It is simply a matter of developers, end users and equipment manufacturers making it a priority. 

“Noise is the easiest engineering problem to solve. I don't know why we're not solving it,” Deering said. “Data center projects are giving whole communities tinnitus.”

Deering also criticized what he called a one-size-fits-all approach to data center design pervading the industry.

While the technology inside data centers has been transformed by the demands of AI computing, their outward appearance has changed little over the past decade. To the untrained eye, they remain massive, drab, warehouselike buildings flanked by rows of generators. 

This uniformity of design made sense when data centers were mainly being built in industrial parks, Deering said. But as AI campuses spread into new communities, these designs became problematic.

Although projects often see cosmetic design tweaks to accommodate local codes or appease residents, they are just iterations on designs fundamentally ill-suited for rural or residential settings. 

The industry needs to rethink data center architecture from the ground up, according to Deering. New building designs need to be introduced that are specifically intended to fit more naturally into rural and residential environments. 

It is simply a matter of resources, Deering said. And the future of Big Tech’s AI build-out hangs in the balance. 

“We're just not going to succeed if the design community is giving us copy-paste designs from the cloud era,” Deering said. 

How quickly the industry embraces these changes could determine whether community opposition intensifies or fades in the years ahead, panelists said.

Although the U.S. is in the middle of an unprecedented data center construction boom, many of the most controversial projects outside traditional hubs are still being built or have yet to begin operations. As a result, there are few real-world examples for communities to evaluate whether fears about noise, emissions and other impacts are justified. 

If data center firms can demonstrate they are reducing their facilities’ impact on nearby communities rather than simply promising tax revenue and economic development, the panelists said local opposition will become less of an obstacle to meeting surging AI demand. 

“In five years, we’ll have more proof of concept around data centers,” Riegle said. “We'll have a lot more operational facilities that data center developers will be able to point to.”