Red alarm level: AI sucker threaten to endanger the energy transition!

Red alarm level: AI sucker threaten to endanger the energy transition!

A new study by the eco-institute on behalf of Greenpeace has published alarming forecasts on the energy consumption of artificial intelligence (AI). According to this, the electricity consumption of AI calculation centers will be eleven times as high by 2030 as in 2023. This dramatic increase could significantly endanger the energy transition and the climate goals. The greenhouse gas emissions, which are currently 29 million tons of CO2 equivalents, are expected to increase to 166 million tons, which could have worrying effects on the climate. AI calculation centers will make up almost half of the global emissions of data centers by 2030, although large technology companies such as Google, Amazon and Microsoft have promised to be climate-neutral by 2030.

Jonathon Niesel from Greenpeace warns that the dependence on fossil energies without parallel expansion of renewable energies leads to an enormous increase in energy requirements. The report "Environmental Impacts of Artificial Intelligence", which evaluated over 95 studies, also shows that the water consumption for the cooling of these data centers from 175 billion liters could be quadrupled to 664 billion liters by 2030 in 2023. In addition, the expansion of data centers and the AI ​​capacities could lead to up to five million tons of additional electronic waste by 2030.

renewable energies and efficiency standards

The Greenpeace study calls for more transparency about the consumption of energy, water and raw materials in AI systems. In order to meet the challenges, efficient standards and labels for data centers and AI applications should be introduced. It is also recommended to integrate data centers into renewable energy and heating networks, but without the expansion of atomic energy to cover the electricity requirement for AI.

In Europe, the energy consumption of data centers for AI applications will probably increase to over 150 terawatt hours per year by 2030, which corresponds to almost a triplement and would make up around five percent of the total European power consumption, currently with two percent. The majority of this energy still come from fossil fuels. Despite the obligations of technology groups to cover their energy requirements through renewable energies, current reports show opposite trends, especially with regard to the increase in greenhouse gas emissions.

technological developments and their influence

An example of these challenges is a new AI data center from Xai, which uses 100,000 NVIDIA H100 GPUs, which consume 700 watts each. This data center needs about 70 megawatts, and in combination with the network transmission, this corresponds to the power of 25 wind turbines. Even if companies like Google and Microsoft make efforts to make their data centers more sustainable, reports such as the Google Environment report from July 2023 show an increase in greenhouse gas emissions by 13 percent to over 14.3 million tons, mainly due to the increased energy consumption of data centers.

Overall, the required massive investment in renewable energies and the expansion of the electricity infrastructure must be addressed urgently in order to cover the increasing energy requirement of AI and at the same time not to lose sight of the climate goals. While AI has the potential to improve the climate balance in various sectors, such as agriculture and disaster protection, the focus on the ecological effects of AI is becoming increasingly important.