In late July, the Trump administration released its long-awaited AI Action Plan, which includes steps to cut environmental requirements and streamline permitting policies to make it easier to build data centers and power infrastructure.
But even with massive deregulation, the fact remains: we have no idea where we’ll find all the energy, water, and grid capacity to meet the enormous speed and scale of the emerging AI revolution.
Recently, experts from the International Energy Agency estimated that electricity use from data centers could more than double in the next five years. By 2030, these facilities could use nearly 9% of all electricity in the United States. Without major investments, this growth will strain our power grid and lead to higher energy bills for everyone.
And it’s not just energy. Globally, by 2027, water consumption from AI alone is estimated to reach the equivalent of more than half the annual water usage of the U.K. Researchers at the University of California, Riverside, estimate that a ChatGPT user session that involves a series of between 5 and 50 prompts or questions can consume up to 500 milliliters of water (about the amount in a 16-ounce bottle). Google used a fifth more water in 2022 compared to 2021 as it ramped up its artificial intelligence work. Microsoft’s water usage increased by 34% over the same period. On top of all this, many communities are protesting or rejecting data center construction due to factors like noise disturbances and limited job-creation benefits.
It’s easy to feel overwhelmed by the scale of the task before us as a nation, especially considering that winning the AI race with China requires overcoming all these constraints, quickly. But the answer isn’t despair, or even just deregulation. We need to innovate. Let’s imagine and build data centers that stack as many functions as possible for sustainability, efficiency, and even social good.
While the most obvious sustainability move for data centers is clean energy, permitting wait times and baseload requirements mean many new constructions are developing “behind-the-meter” power plants with natural gas. Even these systems can transform their environmental impact. Imagine if data centers could capture waste heat and CO₂ and put it to the best possible use—for example, fueling nearby industrial-scale greenhouses that grow fresh, high-quality food. The CO₂ emissions, when introduced into greenhouses, could accelerate photosynthesis, significantly boosting crop yields. Heat captured from servers can also keep greenhouses warm year-round. In other words, you could grow local tomatoes, lettuce, and herbs in the middle of a Northeast or Midwest winter by harnessing data center emissions. Because data centers often choose less populated areas for affordable land and available space, these projects could provide fresh produce to rural food deserts, addressing nutritional gaps and stimulating local economies.
In summer months, when greenhouses require less CO₂, innovative data centers could convert excess emissions into clean hydrogen fuel. Emerging carbon-capture and electrolysis technologies can transform emissions into hydrogen, which could power backup systems, fuel cells, or even local transit. Likewise, organic waste from the greenhouses could be composted or converted into biochar onsite, enriching soils, sequestering carbon, and further contributing to local agriculture. Multiple sustainability functions can be creatively “stacked,” maximizing environmental benefits and turning traditional liabilities into assets.
Data centers also offer major untapped potential for sustainable water solutions. Their expansive, flat roofs—often exceeding 100,000 square feet—are ideal for rainwater harvesting. Just one inch of rain on a 50,000-square-foot roof can yield over 31,000 gallons of water, significantly offsetting cooling demands and reducing dependence on local municipal sources. This harvested rainwater can directly irrigate adjacent greenhouses, creating further efficiencies. Tech giants like Google and Microsoft are already starting to demonstrate the value of this straightforward yet promising approach.
Traditionally, data centers are criticized for providing few long-term jobs. Construction might employ up to 1,500 people temporarily, but ongoing operations usually support only about 50 permanent workers. By integrating greenhouse agriculture and carbon capture, data centers can significantly expand employment opportunities. These integrated campuses could provide apprenticeships, educational programs, and hands-on training in data operations, energy management, sustainable agriculture, and related fields. This approach would promote diverse, long-term job creation and deeper community integration, ensuring more meaningful local benefits.
Right now, we’re embarking on the biggest infrastructure development project in multiple generations. We need to think seriously about the choices we’re making regarding emissions, water, and local economies. We’ve done this kind of big thinking before. When America needed cheap power in the 1930s, we built the Tennessee Valley Authority and strung wires to virtually every farmhouse through the Rural Electrification Act. When commerce demanded speed, we carved the Erie Canal and later laced the continent with the Interstate Highway System. When the Cold War called for a moonshot, we answered with Apollo—turning slide-rule sketches into a lunar landing in less than a decade and achieving scores of technological innovations along the way. Each project looked audacious at the outset. Each rewrote what was possible.
AI infrastructure now demands a leap of similar scale. If we pair data center capacity with on-site microgrids, rain-harvesting roofs, carbon-fed greenhouses, hydrogen production, workforce academies, and other innovations, we can meet the demands of AI without undermining communities or nature.
President Trump’s new AI Action Plan includes some sensible and important steps, including expediting permitting for some data centers and semiconductor fabs as well as new initiatives to boost needed occupations like electricians and HVAC technicians. Yet any comprehensive plan to address the AI challenge needs much more serious attention to questions like energy and water sustainability as well as community resilience. The AI infrastructure race can be a positive opportunity for society, but we need to get creative
Autentifică-te pentru a adăuga comentarii
Alte posturi din acest grup
Forget Cowboy Carter or the Eras tour, the hottest ticket this year is for your favorite podcast.
Content creator tours sold nearly 500% more tickets this year compared to 20
Matthew Williams has slept very little since he learned about Sacha Stone’s plan to build a “sovereign” micronation on 60 acres of land near his home in rural Tennessee. What began as a quic
Let’s be honest: Your phone is a jerk. A loud, demanding, little pocket-size jerk that never stops buzzing, dinging, and begging for your attention. It’s the first thing you see in the
I don’t know if you’ve noticed, but email scams are getting surprisingly sophisticated.
We’ve had a handful of instances here at The Intelligence International Headquarters where we’ve h
Interest in virtual private networks (VPNs) has surged in America and Europe this year. Countries on both sides of the Atlantic have recently enacted new age-verification laws designed to prevent
Instagram’s new location-sharing Map feature is raising privacy concerns among some users, who worry their whereab
