We live in an era marked by unprecedented technological advancements, and among them, artificial intelligence (AI) stands out as one of the most transformative. From systems that optimize medical diagnostics to algorithms that help predict natural disasters or improve energy efficiency, AI has become a key tool for driving human progress.
Moreover, its role in sustainability can be highly positive: for example, AI has been used to monitor deforestation through satellite imagery, boost crop yields through precision agriculture, and even reduce energy consumption in smart buildings by forecasting demand. These applications showcase the enormous potential AI has to help address some of the world’s most urgent environmental challenges.
But did you know that the use of AI also has an impact on the environment?
Although it is often associated with innovation and efficiency, the development and intensive use of artificial intelligence also come with a range of environmental consequences that are not yet fully visible. Below, we’ll explain the main negative effects of AI on the environment, based on scientific data, as well as some emerging solutions aimed at achieving a more sustainable balance.
ENERGY CONSUMPTION AND CARBON EMISSIONS
One of the most obvious impacts of AI’s growth is the high energy consumption required to train and run complex models. For instance, a single training session of a large-scale language model can generate more than 300 tons of carbon dioxide (CO₂), equivalent to over 120 round-trip flights between New York and Beijing.
This consumption multiplies when millions of users interact with models like ChatGPT, virtual assistants, automatic translators, or image generators. Recent studies estimate that by 2025, digital infrastructures related to AI will account for 3.2% of the world’s total carbon emissions—a figure comparable to that of the airline industry.
THE WATER FOOTPRINT: THE HIDDEN COST
Energy use isn’t the only issue. Data centers that enable AI operations also require large quantities of water to cool servers and maintain proper temperatures. Each image generated by an AI model can indirectly involve between 2 and 5 liters of water. In just one week of operation, large generative models can consume over 200 million liters of water.
On a global scale, AI infrastructure could soon surpass the water consumption of entire countries like Denmark. This intensive use of such a limited resource as freshwater raises serious concerns in a context of climate change and recurring droughts.
THE SILENT ISSUE OF ELECTRONIC WASTE
The hardware needed to develop and run AI, such as graphics processing units (GPUs), requires a tremendous amount of rare minerals and heavy metals. The manufacturing, renewal, and disposal of these devices are generating a growing volume of electronic waste.
According to research published in El País, it is projected that electronic waste generated by AI will multiply by 1,000 during this decade, reaching between 1.2 and 5 million tons by 2030. This trend adds pressure to ecosystems where minerals are extracted and to waste management systems, many of which are still not equipped to handle these materials safely.
AI SERVING THE PLANET: POSITIVE APPLICATIONS
Fortunately, it’s not all negative. Numerous projects are demonstrating how artificial intelligence can be an ally of the environment. Some examples include:
- Predictive models that anticipate forest fires by analyzing weather variables and historical patterns.
- Algorithms that optimize logistics routes, reducing CO₂ emissions from transportation.
- AI platforms that analyze ocean data to combat illegal fishing or protect endangered species.
- Neural networks that detect leaks in drinking water systems, preventing millions of liters of waste.
TOWARD SUSTAINABLE AI?
The good news is that initiatives already exist to reduce AI’s environmental impact. Some companies and labs are developing more efficient models that require less data and fewer training cycles. Tools like Eco2AI allow tracking of energy consumption and emissions generated during model training, promoting more responsible AI practices.
Additionally, new cooling methods that rely less on water are being explored, such as liquid immersion systems or locating data centers in colder regions, reducing the need for additional energy to cool servers.
BETWEEN PROMISE AND RESPONSIBILITY
Artificial intelligence is one of the most promising technologies of the 21st century, with the potential to transform every aspect of our lives, including the fight against climate change. But it also carries a significant ecological footprint that cannot be overlooked.
As a society, we have both the opportunity and the responsibility to foster AI that is not only intelligent but also sustainable. Through informed decision-making, proper regulation, and innovation with environmental awareness, we can harness AI’s benefits without mortgaging the planet’s resources.