A few weeks ago, a local developer proposed a new data center facility to support AI in a rural area about 30 miles from my house. The public protest and concern about resource constraints were immediate and unyielding. The global resistance to tech infrastructure just showed up in my backyard.
I felt conflicted. While professionally I am fascinated by the progress of AI, I also consider myself an environmentalist and believe it's our job as humans to correct the damage we've done to the planet, not make it worse.
AI benefits are real
Whatever your views about building new data centers, the capabilities they power are undeniable, and demand is unlikely to slow. A recent SmarterX report found that 74% of professionals see AI as critical or very important to their success in the next 12 months.
Similarly, PwC predicted AI could add $15.7 trillion to the global economy by 2030. Those numbers reflect a real shift: AI already compresses research timelines, automates routine work and creates new capabilities that were impractical five years ago.
Track deforestation in real time using drones and satellites.
Accelerate the discovery of new materials for solar cells and carbon capture.
The same technology that strains the power grid is the one we're counting on to manage it.
The same technology that strains the power grid is the one we're counting on to manage it.
Infrastructure costs affect everyone
That energy infrastructure may be stretched to its limits. Bloom Energy expects U.S. data center demand to nearly double from 80 gigawatts in 2025 to 150 gigawatts by 2028. A single hyperscale facility can use as much power as 100,000 homes, according to the Lincoln Institute of Land Policy. That's significantly driving up costs in some areas.
A Brookings Institution report found that in the Pennsylvania, New Jersey and Maryland grid region, which serves 65 million people, power supply costs jumped from $2.2 billion to $14.7 billion in one year, with data centers accounting for nearly two-thirds of that increase.
Communities are connecting those dots. Pew Research found that roughly 39% of Americans believe data centers have a "mostly bad effect" on the environment, while only 4% said "mostly good."
From May 2024 to June 2025, $18 billion in U.S. data center projects were successfully halted and another $46 billion delayed, according to the latest report from Data Center Watch. Early this month, Monterey Park, Calif., enacted the first municipal ban on new data center construction by an 86% vote. And the backlash is growing as awareness spreads.
For CIOs, access to the power they need is more than an ethics question; it is a supply chain and planning challenge. Data center construction continues to fall behind, and there is no guarantee that the power or telecommunications infrastructure will be able to support them when the buildouts happen.
Enterprise AI strategies must factor uncertainty into their plans and into cost evaluations.
3 questions to ask before you sign off
Despite all the challenges, organizations should not slow down AI adoption. In fact, throwing more AI at these problems may solve them. But CIOs are in a unique position to critically evaluate operational potential vs. infrastructure cost and make the best judgment. Ideally, we can find ways to support AI while using natural resources responsibly and advocating for sustainability.
Before committing to AI infrastructure, whether investing in an AI factory or negotiating with a service provider, IT leaders should ask these questions:
Is my use case likely to succeed? A vague AI initiative with unclear benefits is a different proposition than a targeted deployment with defined outcomes. The more precisely you can build a business case for the problem you're solving and quantify the results, the more honestly you can weigh it against the cost.
Does the expected value justify the resources? Training a large foundation model, running generative AI at consumer scale and deploying a self-hosted inference model require different computing processes that consume vastly different amounts of energy. A complex model that serves only part of your business may not be worth it when you do the math. Get the specifics from vendors and internal teams to understand those details before making any commitment.
Does your AI infrastructure support the local community? Before signing a contract, ask questions: What is the facility's projected energy draw, and does it generate power on-site to offset strain on the local grid? What does water consumption look like relative to availability? Does the facility generate meaningful long-term local employment? After they're built, most data centers employ fewer permanent staff than you might expect.
Also, find out if the operator has a track record of genuinely working with the community. Has it held public meetings, disclosed impact studies and been responsive to local concerns? Some operators are beginning to build facilities with the community in mind, creating multi-purpose spaces with public access and offices. Seek out vendors or contractors that emphasize community engagement and sustainability in their plans.
We don't need to be either a technophile or an environmentalist; we can be both.
Embrace AI with cautious optimism
The data center in my area is not moving forward. When faced with questions from residents, the developer had few answers and withdrew his proposal within the week. I'm sure a similar situation may arise again, possibly much closer to my house. If that happens, I know I'll feel conflicted, but I think that's okay.
We don't need to be either a technophile or an environmentalist; we can be both. We do need to ensure that our technology adoption follows a rigorous process: knowing what we're building toward, understanding what it costs, and asking hard questions of the people building it. The planet doesn't need us to abandon AI. It needs us to proceed with vigilance and responsibility.
Susan Fogarty serves as vice president of Editorial, Enterprise Technology at Informa TechTarget, overseeing a global team producing content for premier IT and telecom media brands including TechTarget.com, InformationWeek and Light Reading.
