The rise of artificial intelligence (AI) is reshaping our world, but it comes with significant environmental concerns. Recent discussions around high-energy consumers like GPT-5 have highlighted the extensive resources these technologies require. Research indicates that GPT-5 alone uses enough electricity to power 1.5 million homes each day, not to mention the enormous amounts of water necessary for data center cooling. However, as technology advances, so too do strategies to mitigate these impacts.
The UK Environment Agency, in partnership with techUK, surveyed 73 data centers across England, revealing promising findings. These results challenge the prevailing notion that AI utilizes vast amounts of water and energy. In fact, 51% of these data centers have adopted waterless cooling systems, and 64% consume less than 10,000 m³ of water annually—equivalent to a small recreational center’s usage. Moreover, 89% actively monitor their water consumption to promote eco-friendly practices.
Richard Thompson, Deputy Director for Water Resources at the UK Environment Agency, expressed encouragement at these findings, emphasizing the need for sustainability in technology. “Advancements must align with safeguarding public water supplies and the environment,” he remarked, further highlighting the importance of innovative cooling technologies.
Energy Use: The Bigger Challenge
Despite the encouraging water usage statistics, the energy consumption of AI data centers remains a pressing issue. As the demand for AI continues to rise, so does the strain on the planet’s energy resources. Even though progress has been made since 2023, the global energy crisis is exacerbated by AI’s voracious appetite for power.
Pinpointing exact energy consumption can be difficult, as many AI companies do not disclose their usage data. A recent study from Berkeley Lab directly correlates energy usage with the increasing popularity of AI assistants, emphasizing the need for better transparency in this sector.
According to Sam Altman, CEO of OpenAI, an average query executed by ChatGPT consumes approximately 0.34 watt-hours of energy. While that may seem small, it significantly adds up with millions of queries each day. Google’s own metrics show an average use of 0.10 watt-hours and 0.12 mL of water per query, though they caution these figures substantially underestimate their total environmental impact.
Are AI Energy Requirements Reducing?
The demand for energy, however, is decreasing as the technology evolves. Google recently revealed that the energy and carbon footprint for a median Gemini text prompt dropped dramatically over a year, illustrating that AI can become more efficient while still improving output quality.
Even as these improvements emerge, Google and other major players must innovate continually to reduce AI’s overall impact. With a significant portion of the American population now utilizing AI technologies daily, the call for solutions that curtail energy use is growing louder.
How Can We Balance AI Growth and Environmental Responsibility?
This key question drives current research and innovation. Tech giants like Google are exploring comprehensive sustainability solutions. Their strategies include scaling back usage during peak times and employing techniques that minimize the workload, thereby reducing energy consumption.
However, even with these initiatives, Google’s energy usage has doubled in the past four years. Nearly 30% of Americans are using AI daily—and many are concerned about the associated energy costs. The challenge lies in finding a balance between innovation and environmental stewardship as the demand for AI continues to grow.
It’s essential for organizations, including OpenAI and Google, to prioritize sustainability as they develop cutting-edge technologies. While the journey of AI technology mirrors the evolution of computers—from room-sized machinery to pocket-sized devices—it must now focus on reducing its resource demands and environmental footprint. The recent discovery that some data centers are using less water than expected is a hopeful sign, yet it represents only a small step in a much larger quest for sustainable AI.
As we forge ahead, we must remain diligent in our commitment to environmental ethics in technology. This means advocating for energy-efficient practices and innovation while continuing to support the evolution of AI. If you want to keep exploring the intersection of technology and sustainability, visit Moyens I/O for more insightful content.