AI Environmental Impact: How Is It Destroying The Environment?

The artificial intelligence revolution is here, and it’s transforming everything from how we work to how we communicate. But under the beautiful surface of ChatGPT conversations and AI-generated images lies a really uncomfortable truth: AI Environmental Impact is becoming one of the most pressing concerns of our digital age. 

As billions of users interact with AI tools daily, the hidden costs of this technology are mounting—and they’re measured in carbon emissions, water consumption, and strained power grids.

Let’s understand more about it in depth.

The Hidden Carbon Cost of AI

Every time you type a simple question into ChatGPT or generate an image with an AI tool, you’re triggering a complex chain of events in massive data centers across the world. These facilities house thousands of powerful processors that consume enormous amounts of electricity, and the numbers are staggering.

Training large AI models like GPT-3 produces approximately 626,000 pounds of carbon dioxide that is equivalent to about 300 round-trip flights between New York and San Francisco, or nearly five times the lifetime emissions of an average car. 

But that’s just the training phase. The real problem begins when millions of people start using these models. Think about the staggering number though.

Each ChatGPT query can emit between 2.5 to 5 grams of CO2, with detailed calculations showing approximately 4.32 grams per query. Compare this to a traditional Google search, which emits just 0.2 grams of CO2, and you can see the problem. When we’re talking about billions of queries per day, these “small” numbers add up fast.

Data Center Energy Consumption: The Power-Hungry Backbone

Data center energy consumption has exploded in recent years, and AI is the primary driver. Researcher Jesse Dodge calculated that one query to ChatGPT uses approximately as much electricity as could light one light bulb for about 20 minutes. 

Multiply that by billions of daily interactions and usage, and you have a serious energy crisis on your hands. Think about it, what’s the future going to be?

The scale is mind-boggling. Goldman Sachs estimates that data centers will be using 8% of total power in the United States by 2030, up from 3% in 2022. 

Before the AI boom, data centers already accounted for 1 to 1.5% of global electricity use and around 1% of the world’s energy-related CO₂ emissions.

But it gets worse. A typical AI data center uses as much electricity as 100,000 households, and the largest facilities under development will consume 20 times more. To put this in perspective, the three major GPU producers (NVIDIA, AMD, and Intel) shipped 3.85 million GPUs to data centers in 2023, up from about 2.67 million in 2022, and that number has only continued to rise.

The Water Crisis Nobody’s Aware About

While carbon emissions grab headlines, water consumption might be AI’s most alarming environmental impact. Google’s hyperscale data centers averaged approximately 550,000 gallons (2.1 million liters) of water per day over the past year. That’s enough to supply thousands of households.

A medium-sized data center can consume up to roughly 110 million gallons of water per year for cooling purposes, equivalent to the annual water usage of approximately 1,000 households. Larger facilities are even thirstier—some can “drink” up to 5 million gallons per day, or about 1.8 billion annually, usage equivalent to a town of 10,000 to 50,000 people.

The human cost is direct and personal. Scientists at the University of California, Riverside estimate that each 100-word AI prompt uses roughly one bottle of water (or 519 milliliters). The AI model GPT-3 is estimated to consume 500 ml of water per 10-50 responses. When multiplied across billions of users, the total water footprint becomes enormous.

By 2027, the situation could become dire. Global AI demand is expected to account for 1.1 to 1.7 trillion gallons (4.2 to 6.6 billion cubic meters) of water withdrawal—more than 4-6 times the total annual water withdrawal of Denmark.

What AI Companies Are Doing Wrong

The tech giants driving the AI boom have made ambitious climate pledges, but their actions tell a different story. Between 2020 and 2023, Microsoft’s disclosed annual emissions increased by around 40%, from the equivalent of 12.2 million tonnes of CO₂ to 17.1 million tonnes. Meta’s Scope 3 emissions increased by over 65% in just two years, from the equivalent of 5 million tonnes of CO₂ in 2020 to 8.4 million tonnes in 2022.

Google’s emissions were almost 50% higher in 2023 than in 2019, and the company has admitted that planned emissions reductions will be difficult “due to increasing energy demands from the greater intensity of AI compute”. Even more concerning, starting in 2023, Google wrote in its sustainability report that it was no longer “maintaining operational carbon neutrality”.

The fundamental problem is the company’s priorities. As researcher Jesse Dodge explains, “Google’s real motivation here is to build the best AI systems that they can, and they’re willing to pour a ton of resources into that, including things like training AI systems on bigger and bigger data centers all the way up to supercomputers, which incurs a tremendous amount of electricity consumption and therefore CO2 emissions”. It’s like capitalism at its peak.

Perhaps most troubling is the lack of transparency. A 2016 report found that fewer than one-third of data center operators track water consumption. 

Technology companies don’t always reveal how much water their data centers use, making it nearly impossible for the public to hold them accountable.

The AI Impact on Environment: Beyond Energy and Water

The AI impact on environment extends far beyond just electricity and water. Let’s look at the full picture of environmental destruction:

Manufacturing Footprint

A GPU’s carbon footprint is compounded by the emissions related to material and product transport, as well as environmental implications of obtaining the raw materials used to fabricate GPUs, which can involve dirty mining procedures and the use of toxic chemicals for processing.

The embodied emissions from constructing data centers are substantial, including concrete, steel, and IT hardware, with Scope 3 GHG emissions for data centers ranging from approximately one-third to two-thirds of overall lifetime emissions. Between 2020 and 2023, Microsoft’s carbon footprint grew by 30%, largely due to the emissions associated with steel, concrete, and chip manufacturing.

The environmental impact of AI is way beyond what you can imagine.

Electronic Waste Crisis

These numbers will shock you.

Among the 62 million tonnes of e-waste produced in 2022, less than one quarter was properly recycled. The annual generation of e-waste is rising by 2.6 million tonnes annually, on track to reach 82 million tonnes by 2030, a steady 33% growth.

The rapid advancement of AI means faster device depreciation, creating even more hazardous waste.

Regional Water Stress

The problem isn’t just the quantity of water—it’s where that water is being taken from. Hundreds of new AI data centers have been built across the US in recent years, most of them in areas already struggling with high levels of water stress. 

On hot days when residents and businesses need water most, data-center water demand spikes too. In Arizona, a data center’s monthly water usage during the summer can be nearly twice its average level.

Long-Term Impact: A Climate Catastrophe in the Making

The trajectory we’re on is unsustainable, and the long-term consequences could be catastrophic. If we don’t take the right step at the right time.

Runaway Emissions Growth

The projected total carbon footprint from the top 20 AI systems in terms of carbon emissions could reach up to 102.6 Mt of CO2 equivalent per year, potentially having a substantial impact on the environmental market, exceeding $10 billion annually when considering potential carbon penalties.

Current estimates of greenhouse gas emissions from AI exceed 300 million tons per year and are likely to grow this decade. As AI models become larger and more complex, these numbers will only increase.

Power Grid Strain and Economic Costs

An analysis by the Union of Concerned Scientists found that in 2024, homes and businesses in Illinois, Maryland, New Jersey, Ohio, Pennsylvania, Virginia and West Virginia faced $4.3 billion in additional costs from transmission projects needed to deliver power to data centers. Regular consumers are already paying the price for AI’s energy appetite.

Water Scarcity and Social Inequality

In 2024, one data center in Iowa consumed 1 billion gallons of water—enough to supply all of Iowa’s residential water for five days. 

As freshwater becomes increasingly less due to climate change, the competition between AI infrastructure and human needs will grow significantly, potentially sparking social unrest and deepening inequalities.

The Existential Risk

As AI researcher Alex Hanna warns, “There’s a lot of people out there that talk about existential risk around AI, about a rogue thing that somehow gets control of nuclear weapons or whatever. That’s not the real existential risk. 

We have an existential crisis right now”. The material infrastructure required to support AI is actively contributing to climate change, the very real existential threat we’re already facing.

How to Mitigate The Risks Of AI?

While the situation is terrible, there are concrete steps we can take to reduce the AI Environmental Impact:

1. Energy Efficiency and Hardware Innovation

Direct water-cooling solutions can recycle loops of warm water to cool data center systems, enabling up to a 40% reduction in power consumption and a 3.5x improvement in thermal efficiencies compared to traditional air-cooled systems. 

Companies must start thinking about this and start taking action.

2. Renewable Energy Commitment

Tech companies must move beyond carbon offsets and actually power their data centers with renewable energy. In water-stressed regions, the priority should be low- to zero-water cooling systems to reduce direct use, while investing to add renewables to the local grids.

3. Model Optimization

Not every task requires the most powerful AI model. General, multi-purpose AI models are orders of magnitude more energy-intensive than task-specific models. By using appropriate models for specific tasks, we can dramatically reduce energy consumption.

4. Transparency and Accountability

The Artificial Intelligence Environmental Impacts Act of 2024, introduced by Massachusetts Senator Edward Markey, mandates that NIST develop standards for assessing AI’s environmental impact and create a voluntary reporting framework for AI developers and operators. This type of regulation needs to become mandatory worldwide.

5. Strategic Location Planning

Building data centers in colder countries can offer a natural cooling system—Facebook built a data center in Luleå, northern Sweden, and Google invested billions into its data center campus in Hamina, Finland, which uses seawater for cooling.

6. Consumer Awareness and Responsibility

Individual users need to be mindful of their AI usage. Ask yourself: do you really need AI for this task, or would a traditional search suffice? Every unnecessary query contributes to the problem.

7. Corporate Accountability

The hype over AI has led many to use these models without questioning whether they run counter to the environmental goals they purportedly will help achieve. Companies must be held accountable for their environmental promises, and penalties must be severe enough to force real change.

The Future

The irony is painful: we’re using massive amounts of resources to build AI systems that we hope will help solve climate change, while simultaneously accelerating environmental destruction. AI applications in end-use sectors could lead to 1,400 Mt of CO2 emissions reductions by 2035 in a Widespread Adoption Case—but only if we deploy AI responsibly.

The AI revolution doesn’t have to be an environmental disaster. With the right regulations, technological improvements, and corporate accountability, we can harness AI’s potential without destroying the planet in the process. But the window for action is closing fast. Every day we delay implementing these solutions, the AI Environmental Impact grows larger and more difficult to reverse.

The choice is completely ours: continue down the current path of unchecked growth and face environmental catastrophe, or demand change now and build a sustainable AI future. And right now, what we’re doing is unsustainable, irresponsible, and potentially catastrophic for our planet’s future.

Frequently Asked Questions