Tools like ChatGPT, application artificial intelligence systems like Photoshop or image generation like Midjourney are ideal for some workflows. However, even those that are free are extremely expensive. A lot of computing power is needed to meet our requests, which translates into a large demand for energy and water to cool data centers.
It is estimated that the use of AI is going to skyrocket global energy consumption (so much so that there are companies like Google or Meta that are going to use nuclear energy to power their needs). However, the International Energy Agency is clear that we are overestimating the consumption of AI. The real problem will be global warming.
More than entire countries. Artificial intelligence has been with us for a long time, but it has never been so accessible and, therefore, in demand as until now. For a long time, we have had access to algorithms that allowed us to streamline processes, but the arrival of this generative AI increased energy consumption by 10. Two of the largest players are Microsoft and Google, which each reported energy consumption of 24 TWh in 2023.
This fact may not say much, but they already consume more energy than 100 different countries. In fact, there are those who pointed out that both – separately – were between Libya and Azerbaijan in energy consumption.
What the IEA says. The International Energy Agency itself has spoken repeatedly about the high consumption of AI – data centers – and how this will multiply in the short term due to this increase in demand for systems based on artificial intelligence. However, energy demand may also be overestimated.
According to the latest global energy outlook report, the IEA stated that, although investment in artificial intelligence increases, the hardware will become increasingly more efficient (more tasks consuming less energy) and, in addition, the energy demand of data centers will be lower than that of other industries. In fact, much less.
Air-conditioning. According to the data managed by the IEA, AI data centers will have an energy demand until 2030 of about 202.8 TWh, exactly what is expected for desalination systems (which also consume a lot of energy to be able to offer drinking water), and far behind other industries such as air conditioning or electric cars. Specifically, the 3% increase in demand expected for data centers is expected to be only a third compared to what will be needed to cool rooms in 2030.
The estimated air conditioning consumption is 676 TWh. And yes, the energy demand of data centers will also be less than 473.2 TWh to heat rooms in the cold months. The IEA comments the following:
“Globally, data centers represent a relatively small proportion of total electricity demand growth through 2030. More frequent and intense heat waves than expected or stricter performance standards applied to new appliances, especially air conditioners , generate considerably larger variations in projected electricity demand than an optimistic scenario for data centers. Rising global temperatures will generate more than 1,200 TWh of additional cooling energy demand worldwide by 2035 in the STEPS, an amount. higher than the electricity consumption of the entire Middle East today.
This is a computing unit of a data center. It is cooled by water and its consumption is considerable
It is still a high consumption. Now, just because the rise of AI is not going to be catastrophic for global energy demand does not mean that it is not very high. So much so that the IEA itself has called a world summit to discuss how to confront the rise of AI. It is something that will be celebrated on December 5 in Paris and the different protagonists of the sector will meet.
Maybe the problem is not—only—energy. On the other hand, and although it is clear that AI is consuming many resources, it is also generating something: a huge amount of electronic waste. Spending on AI is estimated to have increased eight-fold between 2022 and 2023, with much of that money going toward building and equipping data centers.
Furthermore, one of the key points is not only the number of calculation systems, but also their technology. This means that companies are throwing away older equipment in favor of purchasing the latest GPUs from manufacturers like Nvidia, which is generating a huge amount of waste.
It is difficult to reuse a GPU that has been on 24/7 doing calculations, but there are those who suggest that this “old” equipment for demanding tasks be used for others that are not so demanding, such as website hosting, backup copies or your donation to educational centers.
Images | WorldOfSoftware, Garcellor
In WorldOfSoftware | Samsung puts on the handbrake: delays the construction of its cutting-edge factories in the US and South Korea
