By Giannis Kapsalis
The following text is the introduction by Yannis Kapsalis from Greece, to the discussion titled “The invisible cost of AI”, held during the 33rd camp of Antinazi Zone / YRE (Youth Against Racism in Europe), the antifascist and antiracist campaign built around Xekinima (Greek section of ISp). The discussion was also introduced by Leonard Τraistariu from GAS, the Romanian section of ISp and Ζisis Sourlas from Greece.
The rapid advancement of artificial intelligence (AI) is touted as one of the most significant technological breakthroughs of our time, capable of fundamentally transforming work, communication, and the resolution of intractable problems. However, behind the capabilities of these models, lies an unbearable cost: exorbitant energy demands, massive water consumption, and unsustainable pressures on the local communities that host data centers – all within an economic framework of gigantic investments. The issue, then, is not about rejecting technology, but about what purpose it ultimately serves, who pays the environmental price, and who will ultimately reap its benefits.
To start with, we must make it clear that AI technology is a necessary and very useful technological advancement which – if used for the benefit of society and not as a means for major investors and big capital to get richer- can offer many conveniences in daily life and help accelerate other technological achievements. The criticism on AI should have nothing to do with Luddism or the rejection of technological advancement (the Luddites were a movement of 19th-century England, consisting of weavers who rebelled against the newly-built machines of the Industrial Revolution).
For AI models to function, enormous computing power is required, which in turn, requires enormous amounts of electricity. AI companies are building their own data centers so they can meet their own needs and avoid relying on competitors like Google or Amazon, which already have an extensive data center infrastructure. Many of these data centers are being built in the United States, specifically in areas where land is cheap -such as Nevada and Arizona- or in states where AI companies enjoy significant tax breaks.
Environmental Impact
Here are some figures that help illustrate the scale of AI’s needs: At current growth rates, AI will add 100 million metric tons of carbon dioxide to the atmosphere by 2030 – a figure equivalent to the emissions of 10 to 15 million vehicles on U.S. roads. As for electricity, OpenAI’s ChatGPT alone consumes an annual amount of energy equivalent to the annual household electricity needs of nearly three million people in sub-Saharan Africa. The U.S. will consume more electricity for data centers than for the production of aluminum, steel, cement, chemicals, and all other energy-intensive goods combined.
Another natural resource consumed in large quantities by data centers is water, which is used to cool computing systems. It is estimated that annual water consumption for AI needs is expected to reach 731 to 1,125 million cubic meters, an amount equivalent to the household water needs of 6 to 10 million U.S. citizens. It is worth noting that in many areas where data centers have been built, the local community has been left without water, as data centers are being prioritised in the water supply.
The environmental impact of operating AI is enormous. Efforts are being made to pressure AI companies to invest in or rely on renewable energy sources, but in Trump’s America -with its “drill, baby, drill” mentality- breaking free from hydrocarbons is becoming increasingly difficult.
Below are some financial figures related to AI companies. In total, AI companies have invested $1.5 trillion, while their total revenue amounts to just $770 billion. For example, Amazon has spent approximately $273 billion on artificial intelligence infrastructure, while generating only $40 billion in revenue from artificial intelligence. Google’s parent company, Alphabet, has a deficit of $227 billion, having spent $287 billion to earn $60 billion. Meta’s numbers are even worse – $230 billion was spent, and a mere $3 billion was earned.
A Huge Bet by Big Capital
What do these numbers mean? They mean that big capital has placed a huge bet on AI and is trying to benefit at the expense of mining global knowledge. Αll AI models have been developed using personal data and knowledge that all of us have produced without them paying a single cent for it.
Τhe picture that is currently taking shape is already concentrating power at the hands of a few, while shifting the environmental costs and information risks onto the rest of society. On the environmental front, we must demand that data centers consume energy exclusively from renewable sources and are built in areas where the climate or geological conditions are conducive to their construction. We must also demand independent studies and overall planning regarding where data centers will be built, and the final decision should also depend on the consent of local communities.
We can not let AI companies do as they please. The priority must be the benefit to society, not profit. The environmental cost we incur must be returned to society as a multiplied benefit in our lives and be accessible to all people. Especially at this stage, when the benefits of AI beyond the realm of coding are not yet clear, we must be cautious about the sacrifices we are willing to make. In a broader sense, we must demand that these companies are brought under democratic control of their workers and society as a whole, instead of an economic model aimed at the sole enrichment of capital, and open-source AI models that are free and accessible to all.
References:
- United Nations University – ENVIRONMENTAL COST OF AI’S ENERGY USE
- nature.com: Environmental impact and net-zero pathways for sustainable artificial intelligence servers in the USA
- eenews.net: 3-year study maps CO2 spikes from AI data center boom
- isaiprofitable.com
- netzeroinsights.com: The Environmental Cost of Data Centers
- greenpeace.org: The energy and environmental impact of AI and how it undermines democracy
