As part of the latest wave of Big Tech spending on AI, Google recently announced a $9 billion investment to build a new data center campus in Loudoun County, Virginia, part of what's being called "Data Center Alley." This kind of development is not confined to Northern Virginia. Across the country, a data center industrial boom is reshaping landscapes, local economies, and the environment.
The Cool Down set out to map and analyze the impact of these changes across the country, interviewing top experts about whether this growth will pose a threat to local communities or an opportunity to accelerate growth, jobs, and cleaner energy.
Rapidly growing AI data centers: "Power-hungry facilities"
AI Data Centers in the US
There are currently more than 5,400 data centers across the United States. The map above visualizes the vast, nationwide expansion of major facilities directly tied to AI use or adoption that are either already built, expanding, or planned.
Marcus Jecklin, co-founder of Ai4, North America's largest AI industry event, describes the rapid growth of AI data centers as "concentrating large, flexible loads in specific regions, which pressures local grids, water resources, and permitting climate change, but depending on how they're built and powered, they could also be part of the solution."
Energy is required not only to operate these facilities but also to cool their hardware and prevent overheating. Microsoft reported that its water use for AI cooling more than doubled between 2021 and 2022. In Oregon's The Dalles, a Google data center consumed over 274 million gallons of water in a single year.
To meet rising demand, the city secured additional access to the Dog River watershed as part of a $28.5 million agreement with Google. The company transferred industrial groundwater rights to the city in exchange for infrastructure to support its expansion.
Environmental experts warn that if withdrawals continue, the Dog River could run dry for up to eight months annually, threatening aquatic life and fish habitats critical to the region, including salmon, steelhead, and smaller invertebrates essential to the food chain. "With every transformative technology, there are going to be challenges," Sanjay Puri, founder of RegulatingAI and the Chief AI Officer Network, told The Cool Down. "Boundaries that we need to keep. Deciding how we use resources, who uses these systems, and seeing if there is equity in them." Through his work, Puri advocates for innovative AI regulation and works with leaders in business, politics, and civil society to help shape governance frameworks worldwide. "The demand is only going to continue to grow," he said. "… It seems like the demand for AI is limitless."
How to power the AI energy demand - Even as companies tout clean energy pledges, many projects remain tethered to fossil fuel expansion. "If we want to achieve global climate goals, we simply cannot be building out a whole lot of new fossil fuel infrastructure," Webb said. "Yet we're seeing calls to prolong the life of coal plants and build new gas plants just to support data centers."
In Wisconsin, Microsoft is investing $3.3 billion to build a massive campus in Mount Pleasant. To meet demand, state regulators approved two new natural gas plants expected to be run for the next 30 years by We Energies. And Meta's $10 billion project in Richland Parish, Louisiana, will rely on three newly approved natural gas plants to feed its massive power demands. Pollution from emissions linked to natural gas, including nitrogen oxides from burning and methane from leaks; can harm birds, insects, small mammals, and water quality, in addition to human health risks, as covered by the Union of Concerned Scientists.
While many data centers still rely on diesel backup generators, newer facilities are increasingly turning to nuclear power, considered a cleaner form of energy than fossil fuels and there are glimmers of hope showing that fully clean renewable energy can do the job, too.
Amazon Web Services' recent $650 million acquisition of the Cumulus Data Center Campus in Pennsylvania, powered by a nearby nuclear plant is described as "the world's first 24x7 carbon-free, co-located data center." Yet the plant draws water from the Susquehanna River to cool its reactors, returning it warmer in a process that can disrupt aquatic ecosystems, in addition to other nuclear waste and wastewater handling considerations.
But Eco Computing Labs, a Silicon Valley startup, is building off-grid, hydrogen-powered data centers with a mission to deliver zero-emission infrastructure for AI. Its MV1 facility in Mountain View, California (shown on the map), is the first in the world to run entirely on hydrogen fuel cells. Unlike mega-campuses, ECL builds smaller units that still handle heavy AI workloads. MV1 proves an AI data center doesn't need hydroelectric or carbon-based fuels; it can run clean, achieve a negative water footprint, and deliver high performance.
Another option is geothermal power. A recent study published by Project InnerSpace and Future Ventures found that one US geothermal facility could power and cool multiple AI data centers without relying on the grid. It also notes that geothermal is well-suited for the US, since oil and gas workforce skills transfer easily to the industry. In a promising step forward, Google announced a $3 billion deal earlier this year to secure hydro plants in Pennsylvania as part of its goals to power data centers carbon-free.
The Regulatory Gap - The global push for meaningful AI regulation took center stage at the UN's AI for Good summit in Geneva in July. In the United States, however, the federal response is focused more on speed than regulatory safeguards. The AI Action Plan, unveiled the same month as the UN conference, describes AI infrastructure as "critical" and calls for faster permitting, fewer environmental reviews, and fewer regulatory hurdles. "We're seeing enormous pressure to get facilities up and running quickly, often without a full understanding of their environmental and climate impacts," Webb told The Cool Down. "That urgency highlights the need for stronger regulation."
Take AI startup Crusoe, "The AI Factory Company." Working in partnership with OpenAI, Oracle, JPMorgan, SoftBank, and others, Crusoe is planning a $15 billion data center campus in Abilene, Texas, known as Stargate. The joint venture includes up to 85% property tax abatement, eight new buildings, and the creation of over 350 jobs. Under active construction, Stargate is expected to be completed next year. Permit applications have also been filed for a 360-megawatt onsite natural-gas plant to supply power; President Donald Trump has publicly described the project as part of a broader nationwide AI infrastructure rollout.
Corporate sustainability pledges often rely on carbon offsets, renewable energy credits, or promises of emissions-free operations by 2030. These goals may sound ambitious, but they do little to address the immediate reality: the AI revolution is already straining water tables, outdated power grids, and ecosystems. According to the International Energy Agency, data centers in the US are expected to account for nearly half of all electrical demand between now and 2030.
Jecklin suggests, "The cultural shift we need is 'compute-aware development': building with efficiency targets from day one, not as an afterthought. When procurement and engineering share a carbon-and-cost dashboard, efficiency stops being a nice-to-have and becomes a competitive advantage." To do that, he said, "policy is no longer optional. Enterprises need clarity on safety, IP, privacy, and infrastructure. Enforcement is realistic if it's paired with measurement and incentives: standardized reporting, procurement rules that reward additionality, streamlined permits for clean siting, and grid services markets that value flexibility."
For many communities, job creation is the biggest selling point. But the reality is sobering. Once construction ends, data centers require relatively few permanent staff, which can lead to higher utility bills, environmental stress, and far less economic benefit than promised, according to Professor Webb. "The long-term employment benefits for communities are minimal compared to the environmental and public health costs," Webb said. "The reality is that data centers do create great jobs, but they don't create a lot of jobs," Puri added.
The Path Forward - The marvel of AI and quantum computing is undeniable, and their infrastructure is here to stay. OpenAI's ChatGPT alone attracts an estimated 800 million users each week, with 120-190 million logging in daily. As more companies weave AI into their products, the demand for data centers will only grow. The issue is whether regulation and planning can keep pace before the damage becomes irreversible. Puri acknowledges, "Companies are kind of learning and [as they say] flying the plane and building it at the same time." Shifting to renewable, sustainable energy has moved beyond choice; it is now a requirement for AI-powered companies and the communities around them, from Silicon Valley to Data Center Alley, to survive.
But how quickly can that transition happen? And who sets the rules: federal regulators, state commissions, or the local communities left to shoulder the costs? In conversations with leaders of tech companies, Puri observed that developing green AI policies is becoming a higher priority. Securing talent is still their main concern, but there is a greater awareness of the need for renewable power solutions. "The good news is that people are talking about it," Puri said. "Innovation is happening at a very fast pace, innovation that will hopefully create better choices for sustainability, whether it's in AI cooling [methods] or reducing power use."
While the boom continues, community pushbacks are growing. In Bessemer, Alabama, residents and environmental groups pressured officials to pause Project Marvel, a 4.5 million-square-foot data center, after receiving legal and environmental letters, a win driven by organizations like the Southern Environmental Law Center. The message is clear: public scrutiny matters. "Make infrastructure a first-class citizen across the program: utilities, cities, and cloud providers on stage together with enterprises. Bring more community voices, labor, local governments, and environmental groups into citing discussions," Jecklin said.
According to Data Center Watch, $64 billion in US data center projects have been blocked or delayed. The report, published in 2025, identified at least 142 activist groups across 24 states organizing against construction and expansion. "We're going to see more local opposition. Communities want more responsible development, while officials often chase incentives. That disconnect is only going to grow," Webb predicted. When discussing solutions, Webb pointed out, "Community benefits agreements could be powerful tools. They've been used in renewable energy and even stadium projects to ensure local residents actually see some benefits. Data centers should be held to the same standard."
The Alabama case underscores the urgent need for transparent data, stronger local advocacy and meaningful policy engagement. Experts warn the public health burden of data center pollution could reach $20 billion annually by 2030, double that of the coal-steel industry. Yet few systems track or disclose these costs. Communities often lack clear data on how much water or power a facility will consume, or what the long-term effects will be. "One of the biggest challenges is the lack of independent information. Policymakers rely on company data for siting, size, and impacts, which makes it hard to make sound decisions," Webb explained.
The Bessemer case shows that advocacy works. Demanding transparency on energy and water use, tying tax breaks to sustainability, and insisting on public reporting are all ways to hold AI infrastructure accountable. Without reliable data, neither policymakers nor citizens have the tools to respond. "Alignment isn't just a safety spec for models: it's a social contract with each other and the biosphere," Jecklin said. "If we design for sufficiency, transparency, and stewardship, we can earn something closer to abundance with ethics."
Data centers have long powered the digital world, from the music we stream to our latest bank transactions. As we accelerate into an AI-driven future, it's tempting to see the technology as weightless and remote. In reality, its footprint is physical: leaving its trace on power grids, water supplies, household bills, public health, and ecosystems.
The cost of connection is high. And unless regulation and policies grounded in accurate data start guiding expansion, prices will only rise. The demands of training large language models like ChatGPT or Gemini, and running them in real time, have set off a relentless cycle: more servers, more heat, more cooling, more power. Repeat. "These facilities can support important climate research and modeling," Professor Webb emphasized, "but only if they're developed in ways that minimize harm to people and ecosystems. The technology has value; the question is whether we choose to build it responsibly."
Here's how community solar energy can cut your monthly electric bill — no rooftop panels required.
This article is shared with permission at no charge for educational and informational purposes only.
Red Sky Alliance is a Cyber Threat Analysis and Intelligence Service organization. We provide indicators of compromise information via a Notification and a Tier I Mitigation service (RedXray) or an analysis service (CTAC). For questions, comments or assistance, please contact the office directly at 1-844-492-7225, or feedback@redskyalliance.com
Weekly Cyber Intelligence Briefings:
- Reporting: https://www.redskyalliance.org/
- Website: https://www.redskyalliance.com/
- LinkedIn: https://www.linkedin.com/company/64265941
Weekly Cyber Intelligence Briefings:
REDSHORTS - Weekly Cyber Intelligence Briefings
https://register.gotowebinar.com/register/5207428251321676122
Comments