The situation is only going to get worse.
Thirst GPT
China is doubling down on its use of AI, focusing on opening vast new data centers that consume vast amounts of water.
According to a recent report by Hong Kong-based nonprofit organization China Water Risk, the country will soon consume about 343 billion gallons of water in data centers, or the equivalent of the residential water use of 26 million people. It is said that there is a possibility.
By 2030, that amount could rise to 792 billion gallons. This is enough to meet the needs of South Korea's entire population. South China Morning Post I will report it.
down the drain
Training and maintaining AI models is a notoriously energy-intensive and heat-generating task. To prevent data centers from overheating, companies use water to cool their hardware.
According to China Water Risk, China could triple the number of data centers by 2030, reaching about 11 million data center racks housing servers and other equipment.
And it's not just China. The AI boom is already using astronomical amounts of water in the United States and elsewhere. Last year, researchers discovered that OpenAI partner Microsoft's GPT-3 training alone consumed his 185,000 gallons of water, enough to cool an entire nuclear reactor. Google also admitted in its 2023 Environmental Report that it will use up an astronomical 5.6 billion gallons of water in 2022.
AI chatbots rely on specialized chips that consume large amounts of power, so they consume much more energy than traditional methods of searching for information online.
According to a report from China Water Risk, 100 million users chatting on OpenAI's ChatGPT would consume the equivalent of 20 Olympic swimming pools. A simple Google search for the same thing will only say “consume 1 pool”.
Such use could have devastating effects on areas of the world where water resources are already severely scarce. last month, atlantic It was reported that Microsoft was trying to hide water usage at its data center in the Arizona desert.
keep calm and carry on
And the AI hype cycle is far from over. Experts are concerned about further increases in energy and water use.
Rene Haas, CEO of Arm Holdings, said: bloomberg He said this week that he expects the world's data centers to use more electricity than India, the world's most populous country, by the end of 2010.
“We're still at an incredibly early stage in terms of capabilities,” he said, warning that AI is dominating the world's energy needs.
For Haas, one meaningful step forward is finding new ways to train and power these AI models on more energy-efficient chips.
“We need widespread progress,” he said. bloomberg. “Any efficiency is important.”
Learn more about AI and water: Microsoft is draining Arizona's water for AI