In the last few years, the usage of artificial intelligence globally has skyrocketed. According to IEEE Spectrum, ChatGPT alone receives over 2.5 billion prompts per day. In contrast, other popular large language models, such as Claude and Gemini, each receive hundreds of millions of queries per day.
There’s no doubt that AI has been beneficial in many aspects. Large language models like ChatGPT help hundreds of millions, if not billions, of people learn, be more productive, and be more creative. AI has made knowledge more accessible to people around the world, regardless of their background or location. Education barriers that previously existed due to economic disparities are much less of an issue now. As Artificial Intelligence technology improves, it has the potential to better people’s lives and bridge the educational gap between communities.
Due to the ease of access and vast benefits of AI, it has boomed accordingly. Since 2024, ChatGPT’s daily inquiries have more than doubled, and the increased use of AI is rapidly depleting energy and water while emitting enormous amounts of CO2. By 2030, an estimated 24-44 million metric tons of carbon dioxide are expected to be emitted annually from AI datacenters. More than 200 billion gallons of water are needed each year to operate data centers. Much of this water is supplied by local water sources, affecting communities. To reduce costs, AI companies often build data centers in areas with inexpensive land; however, that land is usually cheap because it is located in arid, desert-like regions with limited water resources. People living in these areas are having their drinking water supplies threatened without any say in what AI giants can and can’t do. All of these statistics are only growing as companies are racing to build data centers in response to the AI boom.
On top of environmental repercussions, the surge in demand for materials used to build data centers is driving up the cost of computer components. This isn’t just affecting large companies, but also tech consumers. In the last year, the cost of computer memory (RAM), an essential component, has increased fivefold. Large computer hardware manufacturers are only worsening this issue by the day, rapidly selling out their stock to AI datacenters and then allowing those same datacenters to pre-order stock before it’s made.
The infrastructure necessary to reverse the ramifications of artificial intelligence can’t grow at a rate nearly as close to that of data centers. That said, AI giants need to be more sustainable in how they build and power data centers. Emissions, energy usage, and water consumption are going to spiral out of control if nothing is done now. Without strong efficiency standards, renewable energy integration, and more brilliant model design, the long-term costs of AI could outweigh many of its short-term benefits. Legislation concerning AI’s extensive strain on materials is driving measures such as the Minnesota bill on data center water use, which requires data centers to publicly disclose their total water and energy usage. On top of this, data centers have to pay fees that can run into the millions, on top of the energy they buy, based on how much energy they consume. Governments and technology companies must prioritize sustainable infrastructure and responsible scaling now, or risk turning one of the most powerful innovations in history into a significant environmental liability.