By: Susie
Key takeaways
- AI’s environmental footprint of data centers is extremely damaging. AI data centers increase fossil fuel use, drink up freshwater, generate electronic waste, accelerate covert mining practices, and threaten habitats.
- Due to the pace of AI development and its growing environmental footprint, significant gaps remain in addressing its sustainability. This signals an urgent need to slow its growth to protect our enviornment.
Data centers power our world in so many ways. From cloud computing to banking and healthcare, data centers are subconsciously integrated into our daily routines. It’s safe to say that without them, our modern lives would be impossible.
But another type of data center is brewing. This one is larger, demands more energy and water, and contributes to an ever growing pile of electric waste. Not only do they have significant environmental footprints, the pace of their development leaves little time to plan for the consequences.
Even with environmental considerations aside, some economists claim that AI’s data centers are not profitable investments, either. That is due to the fact that “data center components age rapidly, either [being] made obsolete through rapid advances in technology, or broken down over years of constant, high-powered usage,” says Joe Wilkins with Futurism.
AI might be a promising technology that can generate advances in important fields like medicine, but its reckless development threatens us all. From generating widespread misinformation, creating an even more technologically dependent society amidst a growing need to reconnect to nature, and playing a significant role in global wars, AI is far from its goal of being humanity’s savior.
Indeed, at the Time 100 Impact Dinner which discussed AI’s impact on society, Stuart Russell, co-founder of the International Association for Safe and Ethical AI (IASEAI), said that humans need to “recognize the possibility that not only may the bus of humanity be headed towards a cliff, but the steering wheel is missing and the driver is blindfolded.”
Video provided by YouTube.
AI, data centers, and the environment: What we know
At the heart of the current AI technology boom lies a hidden force: data center development. Once largely absent from the public eye, data centers are now at the forefront of conversations, as communities grapple with new realities in their backyards.
The rapid expansion of AI, or artificial intelligence, has left many people divided. Some are outright opposed to AI, while others find a positive light. But regardless of personal opinions, its explosion has inevitably raised a number of environmental concerns.
So how does the rapid growth of AI, and subsequently, data center development, impact the environment?
It fuels climate change and extreme weather. The rapid development of AI requires a significant increase in the use of fossil fuels, which is a major contributor to climate change. Climate change is a driver of intensifying and extreme weather events.
It strains freshwater resources. Data centers require large amounts of clean, freshwater in order to operate. This places an additional strain on an already struggling resource. Many data center developers do not disclose their water use, creating transparency concerns and leaving many unknowns when it comes to managing water resources.
It accelerates the loss of habitats and open spaces. Data center developments often encompass several hundred acres and threaten ecosystems and open space such as wetlands, streams, forests, and farmland.
It generates large amounts of electronic waste. Computer components that power AI technology have an expiration date and must be discarded and updated in a continuous process, leading to the generation of large volumes of electronic waste that is energy intensive and hazardous to recycle.
It relies on covert global mining operations. AI would not be possible without raw materials from the Earth. Mining is a hazardous and environmentally destructive practice. Without strong regulations in place, it can be detrimental to workers, including children, who work in mines across the globe. Mining operations are often hidden from the public, and their role in AI development is poorly acknowledged.
Researchers who authored a paper on the sustainability of AI conclude that “despite the positive societal benefits, the endless pursuit of achieving higher model quality has led to the exponential scaling of AI with significant energy and environmental footprint implications.” In other words, there may be benefits to AI, but its unrestricted growth may cause more harm than good.
Renewable energy is not enough to power electricity demands for AI
To keep pace with soaring demand, the global electricity consumption is expected to rise in order to power these data centers. Sustainability advocates are hopeful that renewables can power data centers, but some researchers say otherwise. They note that the “infrastructure for carbon-free energy is limited by factors such as geography and available materials, and takes significant economic resources and time to build.”
Due to the mismatch between renewable energy supply and demand for power for data centers, these facilities are increasingly relying on the use of fossil fuels. And as such, their development is revitalizing a dwindling coal era and reigniting the climate change fire.
Authors of an article by MIT Climate and Sustainability Consortium put it well. They say “our capacity to build sustainably cannot keep pace with the data center construction necessary to support Gen-AI, and the global electricity consumption of data centers is expected to continue to rise.”
The sheer volume of electricity required to power data centers is concerning. Some estimates predict that between 2030 and 2035, data centers could account for nearly 20 percent of the global electricity use. And according to researcher Jesse Dodge, a single query to ChatGPT uses as much electricity as powering a lightbulb for 20 minutes.
These estimates demonstrate that AI is an extremely energy intensive technology, and without limits, will strain electric grids and accelerate the growth of fossil fuel use. In order to be truly sustainable, AI cannot continue on its growth path unless it acknowledges its role in fossil fuel consumption.
Freshwater: a resource in decline, and a required component of AI and data center operations
In addition to the demand for electricity, AI’s rapid development has left many communities scrambling for answers on water use. The sophisticated computers housed in data centers generate a lot of heat as they’re working, and large amounts of clean, freshwater is required to cool them.
On average, large data centers can use about 5 million gallons of water per day – the equivalent of a small town with a population of 50,000. Water use is directly proportional to electricity consumption, so the more power a data center uses, the more water it will use. But problems arise because data centers often do not provide actual estimates of water use, leading to inconsistent reporting and essentially gambling with the security of our water resources.
While there are efforts to reduce water waste, such as closed loop cooling systems, the effectiveness is not yet fully understood. In addition, only a handful of data centers may use closed loop cooling systems, or sometimes it may not be disclosed at all what kind of cooling system they use.
Perhaps the biggest concern, though, lies in a two-fold problem. One is that global freshwater supplies are in decline, indicating that there are already significant strains on this resource. NASA satellite images show that the amount of freshwater stored on Earth – in aquifers, rivers, lakes, and wetlands – is on average, declining.
And two is that data centers aren’t legally required to disclose their water use, creating a world of water use unknowns. In addition, prominent data center developers like Google and Amazon have a documented history of deceiving the public and lying about their water use.
Considering that freshwater supplies are already strained, this new era of economic development, which requires even more water as fuel and demands unlimited amounts of it, poses a significant threat to the vitality of our freshwater resources. Until protective measures are in place, allowing data centers to consume freshwater is a risky endeavor.
World map depicting regions that experienced freshwater storage minimums. Image source: NASA.
The lesser known issues: Habitat loss, mining for materials, and electric waste
Electricity consumption and water use dominate data center conversations for good reason, but there are even more problems lingering below the surface. Data centers can span several hundred acres, posing yet another threat to open space like farmland and sensitive ecosystems such as wetlands, streams, and forests. These developments fuel habitat loss, which is the single primary threat to wildlife everywhere.
The loss of habitats also jeopardizes our safety and resilience to climate change and extreme weather. For example, the loss of wetlands increases flooding and reduces water quality. The vast majority of wetlands in Indiana and across the United States have been lost, making it critical we protect the remaining wetlands we can.
On another thread, the mining for materials required to power data centers presents an additional threat to habitat loss on a global scale. Mining is one of the most significant land altering forces on the planet, destroying entire ecosystems and often rendering them unusable for anything beyond restoration. Ecosystem restoration can take several decades, and some ecosystems may never fully be restored to their original condition.
Despite being a serious environmental threat, there is no comprehensive public understanding between the growth of AI and data center development and their link to global mining operations. Mines in other countries besides the United States may have less strict regulations. In some cases, there have been children working in mines, people working without proper safety gear, and large underpaid or unpaid workforces. These covert mining practices undermine transparency, human health, and ecological stability.
And after all is said and done, AI and their data centers have yet another pollution problem: the generation of vast amounts of electronic waste. Electronic waste is one of the fastest growing waste streams in the world. It is considered a toxic and hazardous type of waste due to the variety of chemicals and materials in the electronics. In places where electronic waste is poorly regulated, people and children can become exposed to the harmful pollutants produced by the waste during the recycling process.
What we can do to slow AI’s growth
Clearly, and collectively, AI and data centers are accelerating some big environmental problems. From increasing fossil fuel use, drinking up freshwater, generating electronic waste, accelerating covert mining practices, and threatening habitats, this technology is far from the problem solver it is marketed to be. But there is good news on the horizon: people are paying attention and demanding accountability like never before.
Proactive policy implementation. Among the most important steps to take is slowing the growth of AI. Ideally, this would happen through proactive policy measures aimed at things like disclosing and limiting water and energy use, requiring renewable energy, and preventing habitat loss. In addition, a global mindset is needed to proceed both sustainability and ethically with data center development. Practices like mining and electronic waste recycling are both poorly understood and regulated at a global scale, exposing people and even children to unsafe and unethical conditions.
Research and public education. Additionally, more research and education among the general population is extremely important. An exciting development has emerged to track the environmental impact of data centers thanks to researchers at MIT. Wacuka Ngata, who is studying materials engineering at MIT, grew up in what is known as “Data Center Alley” in Northern Virginia.
Using that first-hand experience growing up near data centers, Ngata is working with other university fellows to develop a database to help further explore the environmental impacts of AI and work towards a more sustainable future for these developments. According to Ngata, “the details that we are capturing in the database will identify specific ecological concerns that will inform next steps for reducing data centers’ environmental footprints.”
Ultimately, the question of whether we can slow the growth of AI and its data centers lies in the hands of all of us. Without engagement, opposition, and education, these developments will continue to grow unchecked. That is why it is important to get involved in any way you can – even if it is simply reading about the impacts that AI will have on the environment.
As data centers come into communities in Indiana and across the world, people are beginning to question their viability and sustainability. Among all the things a community could pour money and resources into – improved roadways, affordable housing, clean water and air, habitat restoration, public transportation – people are waking up to the question: what benefits will a data center actually provide them? The answer to that question remains to be seen.
HEC Data Center Resources
Policy Maker Data Center Guide
Article: Measuring the environmental cost of artificial intelligence and its data centers
Have a comment about this article? Email the author, Susie McGovern, Water Science and Sustainability Specialist at smcgovern@hecweb.org.
Categories: Climate Change, Environmental Justice, Sustainable Development and Green Infrastructure, Water & Wilderness Protection