Congress Faces Backlash Over Fast-Tracking Data Centers as AI’s Environmental Impact Draws Scrutiny

A

coalition

of nearly 120 community, labor, and environmental justice organizations is urging Congress to reject efforts to fast-track data center development tied to artificial intelligence. Their warning reflects a growing tension: the same infrastructure enabling breakthroughs in medicine, education, and scientific discovery is also placing increasing strain on energy systems, water supplies, and local communities.

There is little debate about AI’s potential. It could

accelerate drug discovery

, improve diagnostics, and reshape entire industries. But the physical systems behind that progress are expanding faster than the safeguards designed to manage their impact.

The Scale of the Environmental Impact

The environmental footprint of data centers is no longer abstract. It is measurable, immediate, and growing rapidly.

Large facilities can consume up to

5 million gallons of water per day

, equivalent to the needs of tens of thousands of people. Across the United States, thousands of data centers collectively use billions of gallons annually, and that demand is rising as AI workloads intensify.

This

water use

is directly tied to energy consumption. As servers generate heat, more cooling is required, which increases both electricity demand and water usage. At the same time, freshwater resources are limited, with less than 1% of global water readily available for human use.

The energy side of the equation is equally concerning. Many data centers still rely heavily on fossil fuel-powered grids. Diesel backup generators remain standard, contributing to localized air pollution and long-term emissions. In some regions, these generators operate frequently enough to become a meaningful source of nitrogen oxides and particulate matter, both linked to respiratory illness.

The indirect effects are just as significant. Electricity generation itself consumes water, meaning the true

water footprint of a data center extends far beyond on-site cooling

.

A Growing Strain on Water-Stressed Regions

One of the most troubling trends is where these facilities are being built.

A significant portion of data centers are located in areas already facing water scarcity. Reports indicate that roughly

40% of U.S. data centers operate in regions with high or extreme water stress

, intensifying pressure on local supplies and ecosystems.

In these regions, data center demand competes directly with residential use, agriculture, and long-term water security. Communities have reported rising water costs, increased restrictions, and growing concerns about long-term sustainability.

At the same time, local governments often provide incentives to attract these projects, sometimes without fully accounting for long-term environmental costs.

Pollution, Health Risks, and Rising Costs

Beyond water and energy, the broader environmental impact is becoming clearer.

Fossil fuel-powered data centers contribute to:

• 
  Increased greenhouse gas emissions
  


• 
  Air pollution from backup generators
  


• 
  Higher electricity prices due to grid strain
  


• 
  Infrastructure pressure on already overburdened communities
  

These impacts are not evenly distributed.

Facilities are often concentrated in lower-income areas

, where communities already face environmental challenges. This pattern mirrors earlier industrial expansions, where economic benefits were unevenly shared while environmental costs were localized.

A Contradiction That Is Hard to Defend

The most striking aspect of this issue is not the technology itself, but the mismatch between capability and implementation.

The companies driving the AI boom are among the most valuable in the world, with resources measured in hundreds of billions. They have the ability to deploy cleaner infrastructure at scale. Yet much of the current expansion still relies on legacy approaches that prioritize speed and cost over sustainability.

There is no technical limitation preventing change. In fact, some of the most advanced solutions are already being deployed.

Real-World Examples of Cleaner Data Center Models

Several companies and projects demonstrate that lower-impact data centers are not only possible, but already operational.

Microsoft’s Zero-Water and Immersion Cooling Systems

Microsoft has been experimenting with

two-phase immersion cooling,

where servers are submerged in non-conductive liquid that absorbs heat efficiently. This approach reduces energy consumption and eliminates the need for traditional water-based cooling.

The company has also introduced

next-generation data centers designed to use zero water for cooling

, relying instead on closed-loop systems and advanced thermal management.

In parallel, its

Project Natick

explored underwater data centers, using the ocean’s natural cooling properties to improve efficiency and reduce environmental impact.

Google’s Water Recycling and Efficiency Efforts

Google has implemented

data centers that use reclaimed water instead

of potable water

,

significantly reducing strain on municipal systems. In Georgia, one facility treats and reuses its own wastewater on-site, minimizing external demand.

At the same time, Google has invested heavily in renewable energy, becoming one of the first major cloud providers to

match its electricity use with 100% renewable energy purchases

, setting a benchmark for the industry.

Crusoe’s Closed-Loop Cooling Design

Newer entrants are also pushing boundaries. Crusoe’s large-scale AI data center in Texas uses a

closed-loop cooling system that recirculates water instead of consuming it

, dramatically reducing usage to a fraction of traditional systems.

This approach shows how infrastructure can be designed from the ground up to minimize environmental impact rather than retrofit solutions later.

Startups Advancing Next-Generation Cooling

A wave of startups is focusing specifically on reducing water and energy consumption. Companies like

Submer

and

Corintis

are developing:

• 
  Immersion cooling systems that drastically improve heat transfer
  


• 
  Microfluidic cooling embedded directly into chips
  


• 
  Closed-loop designs that recycle water continuously
  

These technologies are not experimental. They are already being deployed and are expected to scale as demand increases .

Why These Solutions Are Not Scaling Fast Enough

If these alternatives exist, why are they not the default?

The answer comes down to cost, speed, and incentives. Traditional air and water cooling systems are well understood, easier to deploy, and often cheaper in the short term. Newer systems require upfront investment, redesign of infrastructure, and operational changes.

But this short-term thinking comes with long-term consequences. Once a data center is built, it can operate for decades. Retrofitting inefficient systems later is expensive and often impractical.

The Role of Policy in Shaping the Outcome

This is where the current debate in Congress becomes critical.

Fast-tracking permits without environmental review risks locking in outdated infrastructure at precisely the moment when better alternatives are available. The coalition’s argument is not against AI, but against removing the safeguards that ensure it develops responsibly.

Policy can influence outcomes by requiring:

• 
  Transparent reporting of water and energy use
  


• 
  Minimum renewable energy thresholds
  


• 
  Adoption of water-efficient cooling technologies
  


• 
  Community consultation and impact assessments
  


• 
  Tax incentives for renewable energy
  

These measures would not slow innovation. They would ensure that innovation aligns with long-term sustainability.

A Defining Moment for AI Infrastructure

The expansion of data centers is one of the most important infrastructure shifts of the next decade. It will shape not only the future of AI, but also the environmental footprint of the digital economy.

The technology sector has already demonstrated that cleaner, more efficient models are possible. Microsoft, Google, and emerging players are proving that water-efficient and renewable-powered data centers can be built today.

The question is whether these approaches remain exceptions or become the standard.

Congress now faces a decision that goes beyond permitting reform. It is a choice about whether the infrastructure behind AI will reflect the same level of ambition as the technology itself, or continue to externalize its costs onto communities and ecosystems.

The Path Forward: Aligning AI’s Promise With Responsibility

Artificial intelligence has the potential to reshape the world for the better. It can unlock

medical breakthroughs

, optimize supply chains, improve education, and help solve complex global challenges. That future is still within reach.

But turning a blind eye to the environmental and social costs of the infrastructure behind AI is not a viable path forward.

The industry must come together to address these challenges directly. That includes using AI itself to optimize energy systems, improve grid efficiency, reduce emissions, and develop smarter approaches to cooling and water usage. The same intelligence powering innovation can be applied to making that innovation sustainable.

Some

states and local governments are already pushing back

, slowing or rejecting new data center developments due to concerns over water use, energy demand, and community impact. These decisions reflect real pressures being felt on the ground and should not be dismissed.

The responsibility now lies with both policymakers and industry leaders. Instead of removing guardrails, the focus should be on strengthening them. Permitting processes can be improved to be faster and more efficient without sacrificing oversight. Clear, enforceable standards for renewable energy use, water conservation, and community engagement should become the norm rather than the exception.

believes that the future of AI should not come at the expense of the communities it touches. The path forward is not about slowing progress, but about ensuring that progress is built on a foundation that is sustainable, equitable, and aligned with the long-term goal of making the world a better place.

We can and should expect more from the companies building the foundations of this technology. The most profitable industry in the world has the resources to lead, not lag. Progress should come through collaboration, accountability, and higher standards, ensuring AI becomes part of the solution rather than a growing source of the problem.