Chiang Ying-ying/AP Photo
- Nvidia is expanding its partnership with Foxconn to build data centers called "AI factories."
- The aim is for the AI factories to enable development of robots, self-driving cars, and generative AI services.
- The announcement comes as major companies invest in data centers to cash in on the AI hype.
Countries like the US are seeing a factory boom — and so-called "AI factories" could follow.
Earlier this week, Nvidia announced it's expanding its long-standing partnership with Foxconn, the manufacturing giant that famously helps put together the iPhone, to develop a "new class of data centers" that the chip giant said aims to "accelerate the AI industrial revolution."
According to Nvidia, Foxconn's new data centers — which will run on Nvidia's graphic processing units (GPUs) and central processing units known (CPUs) — will help boost development of autonomous machines, such as industrial robots and self-driving cars, and deliver generative AI services akin to OpenAI's ChatGPT.
"A new type of manufacturing has emerged — the production of intelligence," Jensen Huang, the CEO of Nvidia, said during the partnership announcement at Hon Hai Tech Day in Taipei. "And the data centers that produce it are AI factories."
AI factories, Huang said, will be able to support high-powered training efforts that he said will make the AI technology behind its new machines, like self-driving cars, smarter.
"This car would of course go through life experience and collect more data," Huang said during his Taipei announcement, per Reuters. "The data would go to the AI factory. The AI factory would improve the software and update the entire AI fleet."
Nvidia and Foxconn didn't immediately respond to Insider's request for comment when asked where it's building its data centers, when construction will begin, and how much they plan to spend on building them.
The announcement of the AI factories comes as major technology companies like Microsoft invest billions into AI-efforts like building new data centers in an effort to cash in on the generative AI hype.
"There's a well-publicized arms race happening in AI, and the major tech companies are expected to invest $1 trillion over the next five years in this area, mostly to data centers," Jonathan Gray, the president and chief operating officer of investment management company Blackstone, said on an earnings call last July.
Data centers behind AI aren't just expensive. Experts have expressed concerns around the AI industry's environmental impact — which is largely unknown — as they found that data centers require large amounts of energy to run.
Research from the University of Massachusetts found that the amount of energy used to train large AI models is equivalent to the carbon dioxide released for approximately 300 round-trip flights between San Francisco and New York. A recent study on the supply chain of AI devices AI's involving chipmakers like Nvidia found that by 2027, the AI industry could potentially lead to a spike in energy usage that matches the levels produced by countries the size as Sweden and the Netherlands.
Generative AI also requires large amounts of water to cool the data servers behind them. Microsoft was estimated to have used 700,000 liters — or roughly 185,000 gallons — of fresh water to train OpenAI's GPT-3 large language model in its data centers, and Google's latest environmental report shows it consumed the equivalent of 37 golf courses worth of water in 2022, in part, to cool down its servers. One paper even found that ChatGPT servers needs to "drink" the equivalent of a 16.9 ounce water bottle for every 20 to 50 questions it answers.