We scroll, we stream, we ask our digital assistants questions, and with every tap and swipe, we contribute to a digital universe that is expanding at an exponential rate. But behind the seamless cloud services, the instant search results, and the incredible advancements in artificial intelligence lies a dark and often unexamined cost: the environmental toll of the data center. Tech news celebrates the speed and power of these facilities, but it rarely delves into the resource-intensive reality. As AI becomes more sophisticated and our data consumption soars, this silent crisis is accelerating the depletion of our most vital natural resources, and the solutions being touted by tech giants are not as green as they appear.
The AI-Powered Resource Drain
The boom in AI, particularly generative AI, is a primary driver of this escalating resource consumption. The computational power required for these tasks is staggering, and it’s housed in massive data centers packed with thousands of servers that run nonstop. This constant operation generates immense heat, necessitating energy-intensive cooling systems.
And this is where the real cost becomes clear.
Water: Data centers are thirsty giants. Many rely on water-intensive evaporative cooling systems, which can consume millions of gallons of water every day. A single data center can use as much water as a small town. For a more direct example, Google’s data centers consumed a total of 6.1 billion gallons of potable water in 2023, with their Council Bluffs, Iowa facility alone using 980.1 million gallons in that year. As a comparison, this is equivalent to the water consumption of 6.5 golf courses in the southwestern U.S. This places a significant strain on local water resources, particularly in regions already facing drought or water stress. The demand for water for data center cooling is exacerbating the water crisis in water-sensitive regions already suffering from the effects of climate-related water shortages.
Energy: The energy consumption is equally alarming. Data centers worldwide already consume more electricity than many entire countries. The International Energy Agency (IEA) projects that electricity demand from data centers worldwide is set to more than double by 2030, reaching a consumption level of around 945 terawatt-hours (TWh), which would exceed Japan’s current total electricity consumption. In the United States, data centers are on course to account for almost half of the growth in electricity demand between now and 2030, driven largely by AI use. Globally, data center energy consumption in 2022 was 240-340 TWh, which is around 1-1.3% of total electricity demand. This constant energy demand is a problem, especially since more than half of the electricity used to power data centers nationwide in the US comes from fossil fuels.
The Carbon Credit Conundrum
Tech companies are quick to announce their “net-zero” or “carbon-neutral” goals, often touting their use of renewable energy and the purchase of carbon credits. But let’s look behind the curtain.
Carbon Credits: Carbon credits are a system where companies can offset their emissions by investing in projects that reduce greenhouse gases elsewhere, such as reforestation or renewable energy projects. While these initiatives can be beneficial, they are not a silver bullet. They don’t reduce the company’s direct emissions, and they can sometimes be used to justify continued high-emission activities. It’s a way of balancing the books, but it doesn’t solve the core problem of energy and resource consumption. Critics argue that carbon credits can be a form of “greenwashing,” allowing companies to avoid making significant internal emission reductions while still claiming to be “carbon neutral”. As one source notes, if credits are issued to projects that are not “additional” (meaning they would have happened anyway), purchasing those credits will make climate change worse.
The Reality of “100% Renewable Energy”: Companies like Microsoft, Google, and Amazon boast about matching their energy consumption with 100% renewable energy. This is a commendable step, but it’s important to understand the nuance. This often means they purchase a comparable amount of renewable energy available elsewhere on the grid, not that their data centers are running on 100% clean energy at all times. The energy powering the servers in a data center at any given moment might still be coming from a fossil fuel plant, especially in areas where the local grid is not yet fully green. Google has set a more ambitious goal of running on carbon-free energy 24/7 by 2030, a target that would require significant grid-level changes.
The Unchecked Growth of Tech Giants
Despite the environmental concerns, the major players—Microsoft, Google, and Amazon—continue to build new data centers at a breakneck pace. This unchecked expansion is happening without the necessary public and regulatory scrutiny. These facilities are often built in communities that lack the water or energy infrastructure to support them, leading to strained resources. For instance, a Bloomberg report cited by GeekWire noted that nearly two-thirds of the U.S. data centers that were built or are under development in the past three years are located in water-stressed areas. A Google environmental report from 2024 showed that the company’s carbon emissions have risen by 48% since 2019, primarily due to energy consumption from data centers and supply chain emissions, even as the company plans to invest in building even more facilities to meet the demands of its AI tools.
The Reckoning: Is AI Worth the Cost?
While AI offers paradigm-shifting benefits—from medical breakthroughs to climate modeling—we must ask if the proliferation of this technology is worth the environmental cost. Training a single large language model (LLM) like GPT-3 can consume over 1,287 megawatt-hours (MWh) of electricity, which is enough to power over 100 U.S. homes for a year, and emits more than 500 tons of carbon dioxide. While training is a one-time cost, the “inference” stage—when the model is used for real-time queries—also consumes a significant amount of energy, and some estimates suggest it can account for up to 60% of the total energy consumption. A single ChatGPT query uses about 10 times more energy than a standard Google search.
The sheer volume of user queries means this seemingly small energy consumption quickly adds up. As Norman Bashir, a Computing and Climate Impact Fellow at MIT, warns, “The pace at which companies are building new data centers means the bulk of the electricity to power them must come from fossil fuel-based power plants” because renewable energy capacity is not growing fast enough to meet AI’s power requirements.
This environmental cost is not borne equally. Communities most affected by climate change and resource depletion are often not the ones benefiting from AI’s technological advances. For example, while Google’s data center in Finland runs on 97% carbon-free energy, its centers in Asia rely heavily on fossil fuels, contributing to local pollution. The AI Now Institute has highlighted parallels between this uneven environmental impact and historical practices of colonialism. The debate over weighing the costs against the benefits of generative AI is complex and without a clear answer. However, as consumers, we have a role to play in putting pressure on tech companies to make eco-friendly decisions and be more transparent about their energy and water usage.
A Potential Shift: Microsoft’s Recalibration
In a surprising move, Microsoft has recently begun canceling leases on hundreds of megawatts of data center capacity in the U.S. This strategic shift, which affects projects in locations like Wisconsin and Georgia, signals a recalibration of the company’s initial, aggressive investment in AI infrastructure. According to analysts, this is not a retreat from AI but a move to avoid oversupply and optimize resources, especially as the company gains a better understanding of its AI computing needs. Microsoft is reportedly using power delays as a justification to terminate some of these agreements, a tactic previously employed by Meta when it scaled back its metaverse investments.
This move could and should become a trend among other tech giants. With companies like Amazon also pausing some international data center leasing talks, a more cautious and strategic approach to infrastructure growth may be on the horizon. This recalibration allows for a pause in the relentless pace of construction, giving companies time to integrate more efficient cooling technologies and better assess demand. This trend could foster a healthier, more sustainable industry, moving away from a “build-it-and-they-will-come” mentality toward a more measured and environmentally conscious expansion. It’s an opportunity for tech to balance its ambitions with its responsibilities.
Finding the Balance: A Path Forward
Balancing the growing need for data centers with the preservation of our planet is a complex challenge, but it’s not impossible. Here are some steps we need to take:
- Innovate Cooling Systems: Invest heavily in and deploy more efficient cooling technologies. Liquid cooling, for example, can reduce water consumption by a staggering 30 to 50 percent and cut energy demand by 15-20% compared to traditional air cooling. Microsoft is piloting a new chip-level cooling solution that it says will avoid the need for over 125 million liters of water annually per data center. Amazon is also expanding its use of recycled wastewater for cooling, aiming to be “water positive” by 2030 by returning more water to communities than it uses.
- Embrace True Renewable Energy: Companies must move beyond carbon credits and 100% matching to pursue 24/7 carbon-free energy for their data centers. This means investing directly in local renewable energy infrastructure and ensuring their facilities are powered by clean energy around the clock. The IEA urges joint planning efforts, investment in renewables, and smart grid technology to address the challenges of meeting AI-led demand.
- Optimize and Reduce: The industry needs to focus on making AI models and computing more energy-efficient. This includes optimizing code, developing more efficient chips, and utilizing AI to predict and manage energy consumption in real-time.
- Demand Transparency and Scrutiny: We need more transparency from tech companies about their actual resource consumption and emissions. This data should be independently audited and publicly available. Regulators and policymakers need to hold these companies accountable and implement stricter environmental standards for data center construction and operation. As one study notes, almost half of all corporate “net zero” pledges don’t clearly state what emissions they’re addressing, making it difficult to assess their true impact.
- Re-evaluate Consumer Consumption: As consumers, we also have a role to play. We need to be more mindful of our own data consumption and the endless stream of content we demand. Do we need to stream 4K video all the time? Is every piece of data we generate truly necessary to store?
The digital age has brought us incredible convenience and innovation, but it’s time to acknowledge the environmental footprint of our online lives. The dark cost of data centers is a story tech news is only starting to scratch, and it’s a conversation we need to have now, before the bill comes due for us all.
