Efforts like these are under the spotlight as generative artificial intelligence (AI) consumes growing volumes of water, one of the environmental priorities for February's global AI summit in Paris.
The portside data centres "pump water at a rate of about 3,000 cubic metres (100,000 cubic feet) per hour", said Fabrice Coquio, president of their operator Digital Realty's French business.
Water keeping the banks of servers at optimal temperature is "unfit for human consumption... laden with minerals and particles," Coquio added.
Digital Realty counts major platforms, from Microsoft to Disney and TikTok, among clients for its processing and storage.
In recent months the Marseille site has seen installation of a supercomputer powering AI start-up Sesterce's operations.
Every door on the heavily secured site is opened by fingerprint scanner.
Available year-round at a steady 14 degrees Celsius (57 Fahrenheit), the coal mine water enters the server rooms' heat exchangers in a closed-loop system dubbed "river cooling".
The data centre's machines must be kept at 25C and with atmospheric humidity of 60 to 80 percent for top performance, Coquio said.
- 'Different class' -
"If the cooling system stops, any computer room will climb to 45C within ten minutes," potentially halting work, he said.
"Cold is just as vital as the electrical socket powering the machines," Coquio added.
Digital Realty says its river cooling has all but eliminated the need for energy-hungry air conditioning, lowering the three data centres' consumption by 20 percent.
Exploiting such savings will be vital if use of generative AI and its attendant infrastructure grows in line with forecasts.
"The electrical densities our clients are asking for are in a different class than what we've been doing for the last 25 years," Coquio said.
AI chips tend to be more powerful, consuming more energy and producing more heat than traditional processors.
Chipmaker Nvidia's latest hardware is like a "radiator", clocking in at "several hundred watts" of power usage, said Jacques Sainte-Marie, director of the environment programme at France's INRIA computing research institute.
To cool all these chips, many data centres still turn to air conditioning or a water-evaporation system known as adiabatic cooling, in use at Microsoft among others.
- Contested cooling -
Water usage is a major environmental challenge for the AI sector, alongside consumption of power and raw materials.
Google's data centres used 14 percent more water in 2023 than the year before to reach 24 million cubic metres -- enough to fill 9,600 Olympic swimming pools to a depth of two metres (6'6") -- while Microsoft's surged 22 percent.
Alternative cooling methods include ambient air cooling, used by Microsoft in northern Europe and by Digital Realty, French operator Data4 and others.
"Our water strategy is localised and specific to the areas that we operate in, because there's different requirements" depending on the site, said Alistair Speirs, infrastructure chief at Microsoft's Azure cloud-computing arm.
Microsoft is even experimenting with immersing computer components in a coolant that carries heat away but does not damage the parts as water would.
It is not deploying the method widely because "the chemicals that exist to do this right now generally have a high PFAS quotient," Speirs said, referring to so-called "forever chemicals" that have made headlines in recent years as widespread pollutants.
But even methods touted as green alternatives can meet resistance.
In Marseille, the city government and a local campaign group have blasted Digital Realty for monopolising water they say could be used in other ways.
The water "can't be drunk directly, but it could be used to relaunch local agriculture projects, clean streets, water parks," said campaigner Antoine Devillet.
While data centre cooling is getting better, companies could still use that for "greenwashing", he said
"You have to look at the whole production chain, from ores to chips, with huge consumption of electricity, water and sometimes rare raw materials at each step," he said.
Generative AI's environmental impact in figures
Paris (AFP) Feb 1, 2025 -
The surge in generative artificial intelligence (AI) is being met with growing fears about the technology's ecological footprint, one of the top questions up for discussion at a global summit in Paris on February 10-11.
Here are some key figures on the state of play in early 2025:
- Ten times Google's power -
Every request made to OpenAI's chatbot, which is able to generate all kinds of responses to natural-language queries, consumes 2.9 watt-hours of electricity.
That is ten times more than the equivalent figure for a Google search, according to the International Energy Agency (IEA).
OpenAI claims that ChatGPT now has 300 million weekly users making a total of one billion requests every single day.
Beyond ChatGPT, which fronted generative AI's emergence into public consciousness in 2022, there are thousands of chatbots.
One survey by French pollsters Ifop found that 70 percent of 18- to 24-year-olds in the country said they used generative AI.
In America, a Morning Consult poll found that 65 percent of 13- to 17-year-olds used generative AI, with the number close to half for the general population.
- Bigger than France and Germany -
Generative AI would not function without data centres hosting vast reserves of information and computing power.
In 2023, data centres accounted for almost 1.4 percent of global electricity consumption, according to a study by consultancy Deloitte.
But with massive investments planned into generative AI, the figure is expected to reach three percent by 2030 -- or 1,000 terawatt-hours (TWh).
Deloitte said that was comparable with the combined annual consumption of France and Germany.
The IEA forecast a more than 75 percent increase in data centre power consumption by 2026 compared with 2022's levels, to 800 TWh.
American consultancy Gartner said the vast power demands meant that up to 40 percent of data centres built for AI applications could face electricity shortages by 2027.
- Hundreds of flights in CO2 -
Training one of the large language models (LLMs) that power chatbots generates around 300 tonnes of greenhouse gas carbon dioxide, researchers at the University of Massachusetts Amherst estimated in 2019.
That is around the same output as 125 return flights between New York and Beijing.
Two years later, Oxford University researchers put the figure at 224 tonnes for a single training session for OpenAI's GPT-3 model.
Developers have to train thousands of models to push their technology forward.
Despite such estimates, researchers say judging generative AI's overall greenhouse emissions is challenging.
Experts and institutions have pointed to a lack of information on how models are produced, as well as an absence of global measurement standards.
- Rivers of water -
Beyond energy, generative AI also consumes water, especially for cooling computer hardware.
GPT-3 requires around half a litre (one pint) of water to generate between 10 and 50 responses, according to a conservative estimate from researchers at the University of California Riverside and University of Texas at Arlington.
Overall, increased AI demand for water is forecast to amount to between 4.2 billion and 6.6 billion cubic metres (155 billion - 233 billion cubic feet).
That is four to six times the annual water consumption of Denmark, according to the same 2023 study.
- Heaps of electronic waste -
Around 2,600 tonnes of electronic waste such as graphics cards, servers and memory chips emerged from generative AI applications in 2023, according to a study from the Nature Computational Science journal.
The researchers extrapolated that figure to 2.5 million tonnes by 2030 if current trends continue and nothing is done to limit waste.
That would be the equivalent of around 13.3 billion discarded smartphones.
And like much computer hardware, AI equipment including chips requires rare metals to manufacture.
Mining for such metals, often in Africa, can involve heavily polluting processes.
Related Links
Space Technology News - Applications and Research
| Subscribe Free To Our Daily Newsletters |
| Subscribe Free To Our Daily Newsletters |
