Artificial intelligence: innovation vs. energy challenges and environmental impacts

The massive growth of artificial intelligence (AI) is bringing not only technological innovation but also new challenges related to energy consumption and environmental impacts. Scientists are warning about the high energy consumption of AI systems, while tech giants such as Microsoft and Google are reporting significant increases in emissions. Even so, it is not yet entirely clear how much emissions AI will generate in the future, or how much it could potentially save.  

Artificial intelligence (AI) has boomed this year, driving massive investment in the technology. After the initial euphoria, attention is also turning to the technology’s overall effects on various aspects of the economy and society. One question is how much energy it will actually consume in the future, and where that energy will come from.

A recent study by researchers at Cornell University found that generative AI systems such as ChatGPT consume up to 33 times more energy than computers running software designed for specific tasks. Each AI-powered internet query then consumes about 10 times more energy than a standard search. The power required is already reflected in massive investments in data centers around the world.

Given this level of consumption, most analyses assume that the global AI sector will be responsible for 2-4 % of worldwide electricity consumption by 2030. In the United States alone, data centers could consume 9 % of electricity by 2030. That is double their current level.

“AI-powered services involve significantly more computing power – and therefore electricity – than standard online activities, prompting a number of warnings about the technology’s environmental impact,” the BBC recently reported.

As climate policies develop and efforts to mitigate global warming intensify, the question of where the newly required electricity will come from is being raised more and more often. One side points to the ever-growing capacity of renewable sources. Opponents rightly argue, however, that this will not be possible without gas, nuclear and coal sources supplying stable and predictable power to the grid.

Emissions on the rise at Big Tech

Then there is a second question: how much emissions will the required electricity generate? Looking at the latest figures from the technology sector, the answer seems to be quite a lot. Microsoft, which has invested in AI development and placed it at the heart of its product offering, recently announced that its CO2 emissions have risen by almost 30 % since 2020. This is mainly due to the expansion of data centers. At Google, another tech giant, greenhouse gas emissions in 2023 were almost 50 % higher than in 2019. As with Microsoft, data centers are behind the increase.

It should be added, however, that AI’s energy consumption currently accounts for only a fraction of the technology sector’s energy use. According to an analysis cited by the World Economic Forum, the technology sector is estimated to be responsible for approximately 2–3 % of total global emissions. It therefore remains far behind the energy sector and transport.

But this could change quickly in the future if data centers continue to drive rising electricity demand. The analysis also highlights another factor: technology companies depend on supply chains in countries reliant on fossil fuels. Their overall impact on global emissions is therefore likely to be much greater.

It is not yet clear how much innovation will deliver

At face value, this development seems to put AI at odds with sustainability and the fight against climate change. It may indeed do so. The benefits of AI may not be immediately obvious. After all, analysts at Goldman Sachs also expressed doubts about AI’s future benefits in a recent analysis.

Innovation cannot be overlooked today, however, as the potential applications in the energy sector are genuinely wide-ranging. The use of AI could lead to more efficient solutions, for example in grid stabilisation or electricity sharing – areas that directly support the development of renewable sources and decentralisation. These are areas attracting significant investment today not only from multinational energy corporations but also from a growing number of startups.

The US Forbes has listed areas where AI could indirectly reduce emissions. According to the authors, the development of artificial intelligence could help significantly reduce emissions in sectors such as finance and heavy industry, partially offsetting its overall impact on direct and indirect emissions. But we will probably have to wait some time for a more definitive answer on whether AI’s massive expansion will help reduce emissions.