The Impact of AI on Climate Change: Energy Projections and Sustainability
The relationship between artificial intelligence (AI) and climate change is becoming increasingly evident. To better understand this impact, it is crucial to analyze energy demand projections in the context of the growth of advanced AI models. According to the International Energy Agency (IEA), global data center energy consumption is steadily increasing, a trend that will intensify with the expansion of AI and cloud services.
Energy projections in AI development.
Currently, data centers account for approximately 1% of global power consumption. However, with the mass adoption of AI, this figure could increase significantly. A report by OpenAI reveals that the size of AI models doubles every 3-4 months, leading to higher processing and thus energy demands. For example, training language models such as OpenAI’s GPTs requires hundreds of megawatt-hours (MWh), a considerable consumption of electrical resources.
A study by the University of Massachusetts Amherst highlights that training an advanced natural language processing model can generate more than 280 tons of CO₂, equivalent to five times the emissions of a car over its lifetime. If current trends continue, it is estimated that data center energy demand could triple by 2030, especially in regions with rapidly growing technology infrastructures.
Regional impact on energy and water consumption
In the United States, data centers account for more than 2% of national electricity consumption, and this percentage continues to rise as AI and cloud computing take hold. On the other hand, China could lead the way in data center energy demand, driven by the rapid expansion of its AI applications and digital services. This increase also translates into higher water consumption, as centers located in hot areas require intensive cooling systems.
In a scenario of continued growth without energy efficiency improvements, data centers are projected to consume 660 TWh annually by 2040, equivalent to 3% of global electricity consumption. This level of consumption could exacerbate CO₂ emissions, especially in regions where energy infrastructure relies on non-renewable sources.
Towards a sustainable future for AI
Projections make it clear that, without a sustainable strategy, the environmental impact of AI could increase dramatically in the coming years. The challenge lies not only in developing more powerful models, but in doing so in a way that minimizes their carbon footprint. This requires a concerted effort by technology companies, governments and research organizations to ensure that the growth of AI does not compromise global sustainability.
Adopting sustainable approaches, such as the use of renewable energy and improvements in data center efficiency, will be crucial to balancing the advancement of AI with the preservation of the environment.
Modern slavery: a global reality and the key role of trade unions and purpose-driven companies
Modern slavery remains a persistent reality in the 21st century, affecting millions of people around the world. According to the 2023 Global Slavery Index, approximately 50 million people currently...
The urgency of a humanist economy in a dehumanised world, led by Europe
We are living in a time defined by polarisation, extreme individualism, and the pursuit of immediate profit. In this context, the need for a humanist economy is no longer just desirable — it is...
Quantum computing: the brain of the future in service of the planet
Throughout human history, few technological revolutions have promised as much as quantum computing. Often wrapped in mystery due to its scientific complexity, this emerging technology is set to...