Up in the Clouds: How Data Storage Impacts Our Environment

By Pracheth Sanka, Sciences Po—Menton

We live much of our lives online. Whether that is sending an important email via Outlook or Google, uploading pictures of a weekend trip on Instagram, or using large language models like ChatGPT to help with homework, we send and receive millions of terabytes of data on a daily basis. Our electronic data used to be stored locally on hard drives and CDs, but in the mid-2000s, cloud storage gave way to an easier form of data conservation, one that was both cost-effective and easily scalable. Starting in 2006 with Amazon Web Services (AWS), companies began to use large data centers and server rooms. These are warehouses full of hard drives that remotely and electronically store immense amounts of data, providing a service to lessen the load on local hardware and allow for greater data use by clients. Instagram uses AWS, and Gmail uses Google’s cloud storage service, Google Drive, to maintain the data and files of individual users, who can then easily retrieve and access them over the internet, without extra strain on their computers or phones’ storage systems.

Though the cloud facilitates our personal and professional lives, there is a hidden cost to the seemingly passive and mundane requests to recover our data: our environment. Data centers are essentially massive computers, meaning that like the laptops and desktops we use every day, they require enormous amounts of energy to run efficiently as well as an additional amount to remain cool to prevent overheating and hardware damage. Further environmental concerns about land and water usage, chemical pollution, and excess carbon emissions have been raised against cloud storage, but as more and more of our lives rely on this technology, the downsides to data centers have been pushed aside. This year, data storage is expected to surpass 200 zettabytes, or 200 billion terabytes—1 terabyte is four times more than a typical laptop store—therefore additional server rooms and data centers will need to be built to accommodate this workload. Understanding the risks associated with this form of storage can help raise awareness of the undue stress that the cloud places on the environment, and recognizing potential solutions can allow the earth to be unperturbed by our daily online practices.

The Energy Issue

Official reports from the United States Department of Energy (DoE) indicate that data centers consume approximately 4.5% of total electricity consumption in the country—about 176 terawatt-hours worth of energy—and that this figure will likely double by the end of the decade, with estimates showing that consumption could reach 12% of all U.S. electricity use. This energy use comes with a high carbon cost, as nearly 60% of all energy produced in the U.S. still comes from fossil fuels. Some estimates suggest that storing just 100 gigabytes of data, less than half of the 256 gigabytes that the average laptop holds, releases 0.2 tons of carbon dioxide a year. With the vast amount of information held in data centers, the carbon footprint of cloud storage becomes immense. Reports by Morgan Stanley predict that by 2030, cloud storage and data center-related energy production will have released up to 2.5 billion tons of carbon dioxide, a figure that is 40% of current U.S. emissions.

Though some data service companies, like AWS and Microsoft’s Azure, have committed to ambitious net-zero emissions goals, increased demand for cloud storage and cloud computing will require stable and reliable sources of energy. The rapid growth of Artificial Intelligence (AI) and related web services could result in a 12% year-over-year increase in power demand through 2030, according to a report from S&P Global. The report highlights that constraints in renewable energy growth will mean that up to 60% of this new demand will be met by existing fossil fuel-based electricity production, doubling the sector’s carbon footprint to 4% of total U.S. emissions.

While these services do facilitate information storage and computing, storing data locally and avoiding excessive AI use can lessen the large energy consumption of the cloud. Uploading a one-gigabyte document requires about five kilowatt-hours, but downloading it onto a hard drive only requires five milliwatt-hours. That difference is a factor of a million, underscoring the extensive energy required by cloud storage. As these services become continuously necessary, finding energy-saving solutions is vital to reducing their carbon footprint.

Land and Water Use

Along with their extensive energy use, data centers use many natural resources as well, notably land and water. This can have a detrimental effect on local environments by disturbing existing habitats and exacerbating drought conditions, especially in environmentally sensitive areas.

Data centers of the past could operate in typical, small-scale office spaces, but modern data centers often need large swaths of land to establish their commercial operations. Today, average data centers take up two acres of land, and more powerful 40-megawatt centers can be more than seven acres. Large-scale corporations, like Azure and AWS, have projects under construction that will span 500 and 1000 acres respectively. The current largest data center in the U.S. is owned by Meta, spanning 4.6 million square feet. These projects are only becoming larger; the average land acquisition for new centers in 2024 is 224 acres, which was a more than 100% increase from 2022 figures. With broadening expansion, excessive land use is becoming a pertinent issue for data storage and computing. These spaces could be used for habitat preservation, forestry, or other green common-use spaces, such as parks, but are instead used for a polluting and carbon-emitting industry.

Water consumption, however, is a greater cause for concern. Data centers need to use large amounts of water to ensure that their servers and computers are properly cooled. To do this, mid-sized data centers can use up to 300,000 gallons of water a day, roughly the same as the daily use of 1,000 homes; larger-scale operations can use up to three million gallons, which is equivalent to that of a small city. Centers source this water from local watersheds and treatment plants, and since data centers are located throughout the U.S., up to 90% of national watersheds are affected by cloud storage and computing. As the size and scale of centers grow, so does their water use, which is at a detriment to local cities and environments. For example, a Google-operated data center in The Dalles, Oregon, tripled its water usage from 2018 to 2023 and composed a fourth of the city’s water consumption that year.

This increase in use harms already drought-prone areas, especially in the American West. Service operators are attracted to the abundance of cost-effective sources of electricity, namely solar and wind energy, and despite the higher temperatures, opt to pay the relatively cheaper cost of water cooling rather than incur the higher energy cost elsewhere. A study by Virginia Tech shows that one in five centers are located in an area with a moderate or high level of watershed stress. In a state like Arizona, where surface water levels have reached all-time lows, increased consumption by data centers puts further strain on communities that are already reeling from climate-change-related drought, hurting the local ecosystems and cities that need reliable sources of water to survive.

Pollution

Though pollution from data centers is less impactful than their carbon footprint and overuse of natural resources, it is still a significant and harmful side effect of cloud storage. Air pollution from on-site backup diesel generations—which needs to be tested regularly—can release harmful chemicals such as nitrogen dioxide and exhaust into the atmosphere, impacting the health of local animals, including humans. Most data centers use chemical refrigerants to more efficiently cool their systems and computers. These refrigerants can include chemicals that fall under the category of poly-fluoroalkyl substances, better known as PFAS, of which some are considered to be “forever chemicals.” If a leak occurs, some of these chemicals would leach into the ground, contaminating drinking water and affecting soil quality. Others could vaporize and enter the upper atmosphere where they act as greenhouse gasses, further exacerbating global warming. Fortunately, many cooling systems now use PFAS refrigerants that break down into safer compounds when exposed to ultraviolet radiation and many of these refrigerants can be recycled and reused in cooling systems. However, in the case of a leak, these data centers still pose a significant risk to air quality and clean environments due to their use of liquid refrigerants.

Solutions

While the ever-growing demand for data centers may seem like an imminent environmental disaster, several solutions exist to potentially mitigate, or at the least lessen, the impacts of this rapidly expanding sector.

As previously stated, many companies are making strong and concerted efforts to power their data centers with renewable and green energy sources. Major cloud computing companies, like Google and Microsoft, are pledging net-zero emissions by 2030 and 2050, respectively, powered by commitments to full renewable energy use across all company operations, including their data centers.

The construction of centers themselves can be helpful in the fight against carbon emissions and large-scale resource use. Infrastructure within centers, especially related to cooling, can increase energy efficiency to maintain workload with less of a carbon footprint. Examples include direct cooling systems, like Lenovo’s Neptune, which provide direct cooling to system hardware, rather than maintaining the temperature of the entire plant. The physical location of a center is a factor as well, as certain geographical areas facilitate renewable or efficient energy use and are not water-scarce. The aforementioned Virginia Tech study highlights that constructing data centers in the Northeast and Northwest of the U.S. can limit both the carbon emissions footprint and the water scarcity footprint of cloud storage.

While it may be difficult to balance the future growth of the cloud computing and storage sector and environmental considerations, viable solutions exist to limit the harm to our Earth. As the cloud becomes more vital to professional and personal activities, understanding the often unseen and unrecognized effects is increasingly important to push companies to make the requisite change in our collective effort to curb climate change and save the environment.

Cover photo taken from Rawpixel.