Google Vice President Majd Bakar speaks on stage at an annual conference in San Francisco against the backdrop of a giant data center.
Josh Edelson/AFP via Getty Images
hide caption
convert caption
Josh Edelson/AFP via Getty Images
Researcher Jesse Dodge did a rough calculation of how much energy AI chatbots use.
“A single query to ChatGPT uses the same amount of electricity as it would power a light bulb for about 20 minutes,” he said. “So you can imagine with millions of people using something like that every day, that’s going to be really big.”
He is a senior research analyst at the Allen Institute for AI and has studied how artificial intelligence consumes energy. To generate answers, AI uses much more energy than traditional internet uses, such as search queries or cloud storage. According to a Goldman Sachs reportChatGPT queries require nearly 10 times more power than Google search queries.
And as AI becomes more sophisticated, it requires more energy. In the United States, much of that energy comes from burning fossil fuels like coal and gas, which are major contributors to climate change.
Most companies working on AI, including ChatGPT maker OpenAI, don’t disclose their emissions. But last week, Google released a sustainability report with a glimpse into this data. Deep in the 86-page report, Google said its greenhouse gas emissions increased 48% last year from 2019. It attributed the increase to data center energy consumption and supply chain emissions.
“As we integrate AI more into our products, reducing emissions may become more difficult,” the report says.
Google declined to be interviewed by NPR.
“Data centers are getting bigger and bigger all the way up to supercomputers”
Google has set a goal of achieving net zero emissions by 2030. Since 2007, the company has declared its operations carbon neutral thanks to carbon offsets it purchases to match its emissions.
However, starting in 2023, Google wrote in its sustainability report that it will no longer “maintain operational carbon neutrality.” The company said it is still on track to achieve net zero emissions by 2030.
“Google’s real motivation here is to build the best AI systems they can,” Dodge said. “And they’re willing to put a lot of resources into that, including things like training AI systems on increasingly large data centers all the way up to supercomputers, which consumes a lot of electricity and therefore emits CO2.”
Microsoft has taken its climate commitment a step further than Google, vowing to be carbon negative by 2030. However, the company is also facing headwinds due to its focus on AI. In sustainability report In a report released in May, Microsoft said its emissions had increased 29% since 2020 due to the construction of more data centers that are “designed and optimized to support AI workloads.”
“The infrastructure and electricity required for these technologies create new challenges in meeting sustainability commitments across the technology industry,” the report says.
A company spokesman declined to comment further.
AI’s Deep Energy Thirst
AI requires computing power from thousands of servers housed in data centers; and those data centers require huge amounts of electricity to meet that demand.
Northern Virginia has become the epicenter of the booming data center industry. Data centers in this corner of the state will need enough energy to power 6 million homes by 2030, according to Washington Post.
Demand for electricity nationwide has become so great that plans to decommission some coal-fired power plants have been delayed, the report said. another report by the Washington Post.
“There’s a whole physical infrastructure that needs to be built to support AI,” said Alex Hanna, research director at the Institute for Distributed AI. She worked on Google’s ethical AI team but left the company in 2022 after she edited a research paper highlighting the environmental costs of AI.
The data center boom will continue to grow “as long as there are organizations committed to fully investing in AI,” Hanna said.
Goldman Sachs has Research the projected growth of data centers in the United States and are estimated to use 8% of the country’s total electricity by 2030, up from 3% in 2022. The company analysts say “The rapid growth of AI technology and the data centers needed to deliver it” will drive a surge in electricity demand “at levels not seen in a generation.”
Today, there are more than 7,000 data centers worldwide, according to Bloomberg. That’s up from 3,600 in 2015. Combined, Bloomberg estimates these data centers consume as much electricity each year as the entire country of Italy.
The “AI-first” world
All the big tech companies are going all-out on AI. Alphabet CEO Sundar Pichai has called Google an “AI-first” company. Over the past few months, the company has released its Gemini chatbot to the world. and in addition AI Overview Tool for Google Search. Facebook’s parent company Meta has added chatbots to some of its products. And Apple announced a partnership with OpenAI last month to bring AI to its digital assistant Siri.
In their first-quarter earnings, all of these companies said they investing billions of dollars in AI.
Google said it spent $12 billion on capital expenditures in that quarter alone, “primarily” investing in data centers to fuel its AI efforts. The company said it expects to maintain that level of spending throughout the year.
Hanna, the AI researcher, says the environmental costs of artificial intelligence will only get worse without serious intervention.
“There are a lot of people out there talking about existential risk around AI, about some rogue thing that somehow gets control of nuclear weapons or whatever,” Hanna said. “That’s not the real existential risk. We have an existential crisis right now. It’s called climate change, and AI is clearly making it worse.”
Editor’s Note: Google and Microsoft are among NPR’s sponsors.
