• 3 minutes e-car sales collapse
  • 6 minutes America Is Exceptional in Its Political Divide
  • 11 minutes Perovskites, a ‘dirt cheap’ alternative to silicon, just got a lot more efficient
  • 13 hours Could Someone Give Me Insights on the Future of Renewable Energy?
  • 14 hours The United States produced more crude oil than any nation, at any time.
  • 14 hours How Far Have We Really Gotten With Alternative Energy
  • 13 hours "What’s In Store For Europe In 2023?" By the CIA (aka RFE/RL as a ruse to deceive readers)
  • 1 day Bankruptcy in the Industry
Haley Zaremba

Haley Zaremba

Haley Zaremba is a writer and journalist based in Mexico City. She has extensive experience writing and editing environmental features, travel pieces, local news in the…

More Info

Premium Content

Can Tech Keep Up With AI's Appetite for Energy?

  • The tech sector's carbon footprint is rising, with AI's energy demand already comparing to the consumption of developed countries.
  • Google's Alphabet plans to tackle this issue by sourcing energy from clean energy hotspots around the world in real-time, aiming for 24/7 renewable energy usage.
  • Challenges include navigating data sovereignty policies and ensuring grid stability amidst fluctuating demand, highlighting the complexity of achieving sustainable energy consumption in the digital age.
AI Energy

We tend to imagine the internet as something intangible. Data as ones and zeros that float through the cloud. Artificial Intelligence as a sinister savior of an electric pulse. But in reality, the digital age depends on lots and lots of physical infrastructure, from massive cables running alone the ocean floor to warehouse upon warehouse of computer servers. And all that infrastructure requires a massive amount of energy. And as the demand for digital services increases rapidly, so too does the sector’s already prodigious energy consumption and associated carbon footprint. 

This puts the tech industry in a bind. The “hyperscalers” of big tech have made some of the biggest decarbonization commitments of any business in any sector, but they’re struggling to get their consumption under control and the feasibility of those pledges is seriously under threat.  “Currently, the entire IT industry is responsible for around 2 percent of global CO2 emissions,” Science Alert reported last year. And that number is growing rapidly, and will skyrocket in coming years. Already, according to figures from the International Energy Agency, data centers and transmission networks are jointly emitting as much carbon dioxide every year as the entire nation of Brazil.

AI in particular is a massive energy suck, and its recent rapid expansion has really thrown a wrench into the tech sector’s decarbonization plans. Already, AI alone consumes as much energy as some entire developed countries and nearly as much as Bitcoin. Consulting firm Gartner projects that in a business-as-usual scenario, the AI sector alone will be responsible for a whopping 3.5 percent of global electricity consumption by just 2030. It’s estimated that the training process for GPT-3, the predecessor of ChatGPT, used about 1,287 megawatt hours of electricity and 10,000 computer chips. That enough energy to power approximately 121 homes in the United States for an entire year, and enough energy to produce about 550 tonnes of carbon dioxide. 

Related: Europe’s Secret Weapon In Its Energy War With Russia

While tech giants have invested heavily in renewable energy consumption, the growth of AI’s energy demand has far outstripped the growth of clean energy sources able to power it. In response, companies like Google’s Alphabet are employing some out-of-the-box thinking to try to get out ahead of the problem before they miss their emissions targets by a mile. The plan is to completely redesign the way that data centers source their energy, changing over from a steady and singular grid connection to a flexible, hourly grid connection. 

Currently, data centers operate more or less at the whim of the grid they are connected to, and their emissions are a reflection of the local grid’s energy mix, which are often heavily reliant on fossil fuels. But Alphabet is working on software which would be able to locate, on an hourly basis, which grids in which locations are producing surplus clean energy in real time. Data centers would then draw their energy from that grid before moving onto the next clean energy hotspot as needed, essentially chasing the sun around the globe as it shines upon solar panels in each consecutive time zone. In this way, data centers could use 100% clean energy throughout the day. 

Related: Forgotten Gas Reserves Could Be A Gamechanger For European Energy

This strategy could also relieve some of the operational costs of these companies, as surplus wind and solar power is sold at a discount on many markets. This could represent a considerable amount of savings considering how much power tech is drawing at any given time. To paint a picture of just hoe much, it’s estimated that Open.AI, the creators of ChatGPT, spend about US$700,000 per day on computing costs alone in to operate its chatbot for approximately 100 million users worldwide. 

While Google is already piloting this approach, no one has yet achieved the goal of 24/7 clean energy. It’s a promising idea, but will almost certainly be “complicated by countries pushing for data sovereignty policies that attempt to restrict and safeguard the flow of data across border,” as Fortune recently reported. And shifting massive amounts of demand to diffent grids on the hour could cause some serious grid stress if not managed properly.

“Being able to pursue spare zero-carbon megawatts can help reduce stress on grids, such as during a heat wave or frigid winter storm,” Fortune reports. “But data centers need to cooperate with utilities and grid operators because big swings in demand can throw electric systems into disarray, boosting the odds of blackouts.”


By Haley Zaremba for Oilprice.com

More Top Reads From Oilprice.com:

Download The Free Oilprice App Today

Back to homepage

Leave a comment

Leave a comment

EXXON Mobil -0.35
Open57.81 Trading Vol.6.96M Previous Vol.241.7B
BUY 57.15
Sell 57.00
Oilprice - The No. 1 Source for Oil & Energy News