The International Energy Agency has a stark warning: electricity demand from data centers could more than double by 2030, driven overwhelmingly by the explosive growth of generative artificial intelligence. Globally, data centers already consumed an estimated 415 terawatt-hours in 2024—roughly 1.5 percent of total demand—but AI workloads and expanding cloud services are poised to push that figure far higher, according to the IEA’s 2025 Energy and AI report. Nowhere is this surge more visible than in Texas, which has emerged as a magnet for massive new server farms, thanks to its pro-business climate, relatively light regulation, and existing energy infrastructure.

Detailed in a new analysis from Lamar University Press, the physical footprint of AI is drawing fresh scrutiny from environmental engineers, utility planners, and everyday technology users who may not realize that every Copilot prompt, ChatGPT query, or Google Gemini search has a tangible cost in electricity, water, and hardware. Thinesh Selvaratnam, a Lamar University associate professor of civil and environmental engineering, underscored that the issue is not the AI software itself, but the vast constellation of data centers, cooling systems, and power lines needed to keep it running.

The Surge in Power Demand: Data Centers by the Numbers

The IEA’s 2025 report projects that global data center electricity consumption could reach 830 terawatt-hours or more by 2030—more than the entire current power demand of some industrialized nations. In the United States alone, data centers are expected to account for nearly half of all electricity-demand growth between now and the end of the decade. That concentration creates significant strain on local grids, even when the raw global percentage seems manageable.

Selvaratnam pointed out that the sheer scale of modern facilities—some spanning hundreds of thousands of square feet—requires not only massive amounts of energy to power servers but also elaborate cooling infrastructure to prevent overheating. “All those data centers need cooling systems,” he said. “They need electricity, and they need water, and then they eventually need to replace their equipment.”

Why Texas Is Ground Zero for the Boom

Texas currently hosts 411 data centers, second only to one other state, according to Data Center Map. Bloom Energy estimates that Texas will become the nation’s leader within three years, with its share of the data center market set to increase by 142 percent. Selvaratnam attributed this to a combination of factors: no state income tax, a regulatory framework friendlier to industry than many competitors, and an existing infrastructure already geared toward large-scale industrial projects.

“The economic climate in Texas is supportive to industry in terms of taxation and looser regulations,” he explained. “The state-level government is pro-business, and the existing infrastructure is pro-industry.” That has turned cities like Austin into a “next Silicon Valley” for data center development, but it also raises pressing questions about whether utilities and local governments can coordinate enough power, water, and transmission capacity before new facilities come online.

Water, Cooling, and E-Waste: The Broader Environmental Footprint

Beyond electricity, AI data centers exact a toll on water resources. Many facilities rely on water-cooled systems to dissipate the enormous heat generated by rows of servers and specialized AI accelerators. The exact impact varies by facility design and local climate, but in water-stressed regions, even moderate consumption can exacerbate shortages.

Then there is the growing mountain of electronic waste. As AI demand accelerates hardware refresh cycles, older servers, networking gear, and specialized chips are retired at an increasing clip. The UN Environment Programme has long cautioned that improperly handled e-waste can leach hazardous materials—lead, mercury, and other heavy metals—into soil and water. Selvaratnam noted that once these substances enter the water supply, the problem cascades: “It’s not just an environmental problem. As soon as they enter into any of the streams, they become an economic burden.” Health impacts can lead to missed work and school, hitting communities hard.

What It Means for Windows Users and IT Administrators

For the typical Windows user, AI features like Microsoft Copilot appear as seamless software conveniences—a sidebar that summarizes documents, writes code, or generates meeting notes. But every interaction ultimately relies on cloud infrastructure humming in data centers around the world. The energy and water consumed by that infrastructure are invisible to the end user, yet they are part of the true cost of generative AI.

For IT administrators and procurement managers, the implications are more immediate. Choosing cloud providers or on-premises solutions increasingly means weighing not just performance and price but also environmental impact. Microsoft has published sustainability commitments, but the granularity of reporting varies. Selvaratnam’s research advocates for clear disclosure of electricity and water demand, better coordination with utilities, and stronger standards for hardware end-of-life management. Admins who oversee hybrid environments or manage device fleets can push suppliers for transparency and factor sustainability into purchasing decisions.

Developers building AI-driven Windows applications can also play a role by optimizing models for efficiency, using smaller or quantized versions that demand less compute, and leveraging edge processing where feasible. As Selvaratnam put it, “Texas doesn’t need to choose between innovation and environmental stewardship. We can make smart planning so our innovation and stewardship matches.”

How AI Went from Niche to Necessary

Just a few years ago, generative AI was a curiosity limited to research labs and early adopters. Today, it is embedded into Microsoft 365, Windows, and the Bing search engine. An AI overview now appears at the top of many Google searches. Copilot is a default presence in the Windows 11 taskbar. This rapid integration means that the infrastructure behind AI—once the concern of only hyperscale cloud engineers—is now directly linked to the daily workflows of millions.

Microsoft itself has explained that generative AI systems are “built upon the foundation of neural networks” and use advanced machine learning to analyze vast datasets and produce original content. The company notes that breakthroughs in natural language processing, exemplified by ChatGPT, have made interactions more fluid and useful. But that progress has come with a voracious appetite for computing power. Training and running large language models require thousands of specialized GPUs or TPUs running around the clock, and the trend shows no sign of slowing.

Steps You Can Take Today

The problem is systemic, but individuals and organizations can act:

  • Advocate for transparency: Urge cloud providers and hardware vendors to publish detailed per-region metrics on electricity, water, and e-waste. Support legislation that mandates such disclosures.
  • Choose efficient cloud services: When evaluating providers, look for those with strong sustainability reporting and commitments to carbon-neutral or water-positive operations. Microsoft’s Environmental Sustainability Report can be a starting point.
  • Optimize AI usage: Use AI features mindfully. For simple tasks, consider whether a traditional search or local tool might suffice, reducing the load on distant data centers.
  • Recycle electronics responsibly: Ensure old laptops, servers, and networking gear end up in certified e-waste recycling programs to prevent hazardous materials from leaking into the environment.
  • Support smart policy: Urban and regional planning that ties incentives for data center construction to environmental benchmarks can help states like Texas balance growth with stewardship.

Selvaratnam emphasized that incentives should be tied to environmental performance, not just broad tax breaks. “Give the incentives that tie to environmental things,” he said, such as using reclaimed water for cooling or adopting high-efficiency infrastructure designs.

What’s Next for AI Infrastructure

Looking ahead, the tension between AI’s promise and its environmental price tag will only intensify. The IEA’s projections may prove conservative if new AI applications outpace efficiency gains. Tech companies are investing heavily in renewable energy and advanced cooling techniques, but these efforts must scale rapidly to keep up with demand. For Windows users and IT decision-makers, staying informed and demanding accountability will be crucial. As the Lamar University analysis makes clear, AI is not going away—but how we power it is a choice we all have a stake in.