Artificial Intelligence has become one of the main focuses of debate due to its supposedly large energy consumption. According to data from the AMETIC Observatory, training models like Llama 3 consumes as much electricity as more than 1,000 homes in a year. Furthermore, the process generates around 500 tons of CO₂, a carbon footprint equivalent to a round-trip flight from Madrid to New York taken 300 times.
The expansion of AI shows no signs of stopping. If the trend continues, data centers could consume as much energy as an entire country, such as Ireland, by 2026, according to the International Energy Agency (IEA). This raises serious questions about the sustainability of technological advances.
What does this increasing energy consumption mean for our future? Read on to discover the real impact of AI on the energy system.
Why does Artificial Intelligence consume so much energy?
AI's enormous electricity consumption is no coincidence. There are several fundamental reasons:
Increasingly larger models: Each generation of AI has more parameters. GPT-2 had 1.5 billion parameters; GPT-4 exceeds 170 billion. This exponential growth requires more processing hours and greater computational resources.
Intensive training: Training an AI model can take weeks or months, using thousands of high-performance GPUs running 24/7.
Continuous operation: After training, models continue to consume energy to provide real-time answers, analysis, and predictions.
Supporting infrastructure: Data centers require not only computing power but also cooling, security, and redundancy, which further increases energy consumption.