Artificial intelligence is expected to have the most impact on practically everything since the advent of the internet. Wall Streetsure thinks so. The tech-heavy Nasdaq (^IXIC) is up 26% year to date thanks to the frenzy over AI-related stocks.
But AI's big breakout comes at a cost: much more energy.
Take for example OpenAI's chatbot ChatGPT. Research done at the University of Washington shows that hundreds of millions of queries on ChatGPT can cost around 1 gigawatt-hour a day, or the equivalent energy consumed by 33,000 US households.
"The energy consumption of something like ChatGPT inquiry compared to some inquiry on your email, for example, is going to be probably 10 to 100 times more power hungry,” Professor of electrical and computer engineering Sajjad Moazeni told Yahoo Finance.
Industry participants say this is only the very beginning of what's to come.
“We’re maybe at 1% of where the AI adoption will be in the next two to three years,” said Arijit Sengupta, founder and CEO of Aible, an enterprise AI solution company. “The world is actually headed for a really bad energy crisis because of AI unless we fix a few things.”
An energy-hungry Facebook data center under construction. (George Frey/Getty Images) ·George Frey via Getty Images
Data centers are the heart of the advanced computing process. They are the physical locations with thousands of processing units and servers at the core of the cloud computing industry largely managed by Google, Microsoft, and Amazon.
"As you think of this shift towards these larger foundation models, at the end of the day you’re going to need these data centers to require a lot more energy as a whole," Angelo Zino, VP and senior equity analyst at CFRA Research, told Yahoo Finance.
Data centers have increasingly shifted from using simpler processors, called CPUs, to more advanced graphics processing units, or GPUs. Those components, made by companies like Nvidia (NVDA), are the most energy intensive.
"For the next decade, GPUs are going to be the core of AI infrastructure. And GPUs consume 10 to 15 times the amount of power per processing cycle than CPUs do. They’re very energy intensive,” explained Brady Brim-Deforest, CEO of Formula Monks, an AI technology consulting company.
Added Brim-Deforest: "Energy consumption is going to dramatically increase on a global scale, simply because of the energy-intensive nature of AI. But if you look at the nuances, what's interesting is AI is also incredibly efficient at things that humans are not as efficient at."
'Huge massive infrastructure cost'
Research done by Benjamin C. Lee, professor of electrical engineering and computer science at the University of Pennsylvania, and professor David Brooks of Harvard showed that data center energy usage grew 25% a year on average between 2015 and 2021. This was before generative AI grabbed national headlines and ChatGPT usage skyrocketed.
Meanwhile, US Energy Information Administration data revealed an annual growth rate in renewable deployment of 7% during the same period, though that number is expected to increase with initiatives like the Inflation Reduction Act.
“There’s already this fairly large gap between the growth rates, between data center energy, and renewable energy deployments,” Lee told Yahoo Finance.
“We call it cloud computing; it feels like there’s no cost associated with it,” said Lee. “There’s a huge massive infrastructure cost.”
To counteract such consumption, the major cloud providers like Google Cloud, Microsoft Azure, and Amazon Web Services all invest in renewable energy to match their annual electricity consumption. They hold net-zero pledges, meaning they remove as much carbon as they emit.
"Net zero doesn’t mean you’re carbon-free potentially. There will be hours of the day where you don’t have enough sun or enough wind, but you’re still going to be drawing energy straight from the grid at whatever mix the grid will provide to you," said Lee.
He is studying ways in which data centers can shift or reschedule their computation based on the availability of carbon-free energy.
"Maybe you compute much more in the middle of the day when there's a lot of solar energy and you compute much less in the middle of the night," said Lee.
It's no surprise, then, that the energy usage challenge has created a crop of companies aimed at creating more efficient ways of using AI models.
"We can literally cut down energy use ... in these [types] of AI workloads by one-third by just moving to serverless," Aible's Sengupta told Yahoo Finance, describing a technology that uses server resources on demand for more efficiency.
Analysts point out that reducing costs will naturally drive the industry towards energy solutions.
“Whether it’s because of emissions or financial efficiencies or investor pressure or anything else, we do see companies looking more at how to be more efficient. It’s an operational cost. The more efficient you are, the lower your operational cost,” says KPMG's US climate data and technology leader Tegan Keele.
As Zino of CFRA Research pointed out, the winners in the space are the data center operators. "The actual data usage and how all this comes together is going to be more concentrated to a couple of companies out there," said Zino.
"More and more companies are essentially renting space in the cloud rather than kind of investing and building their own data centers, just because in the future I think it's going to be a lot more costly in nature," he added.
Ines Ferre is a senior business reporter for Yahoo Finance. Follow her on Twitter at @ines_ferre.
