Warning on AI Competition and Climate Change Response Issues
Concerns have been raised that major US big tech companies may fail to keep their climate change commitments amid competition in artificial intelligence (AI).
On the 12th, Yonhap News, citing Bloomberg and others, reported that Amazon and Google recently released their annual environmental reports, which included statements that achieving long-term climate goals could become more difficult due to AI business activities.
Previously, the emergence of generative AI that speaks and responds like humans was pointed out as a new variable threatening 'carbon neutrality.' The process of storing data and running complex algorithms requires more electricity than simply obtaining answers through basic searches.
In fact, Google announced last year that its greenhouse gas emissions reached 14.3 million tons, a 13% increase from the previous year. It explained, "As AI-related businesses grow, reliance on data centers that consume massive amounts of energy has increased, resulting in a 48% rise in greenhouse gas emissions over five years."
The process critical to training and operating generative AI models takes place in 'data centers.' Data centers house all the equipment necessary for IT service provision, including server computers, network lines, and data storage devices, in a single building that requires power 24/7. Some data centers contain thousands of Nvidia AI chips, each consuming 700 watts of power. This is eight times the power consumption of a typical 60-inch flat-screen TV.
Google stated that achieving its goal of carbon neutrality by 2030 will not be easy, adding, "There is uncertainty because the impact of AI on the future environment is complex and difficult to predict."
Microsoft (MS) revealed in May that carbon emissions have increased by about one-third since 2020 due to data center construction. MS said, "It has become more difficult to achieve the goal of not only carbon neutrality but also net negative emissions by 2030."
Recently, MS announced the ‘Stargate’ project, planning to invest $100 billion (approximately 135 trillion won) by 2028 to build massive data centers in partnership with OpenAI.
Amazon, which aims to achieve carbon neutrality by 2040, stated, "Although we reduced carbon emissions by 3% last year, we expect challenges related to AI going forward." Amazon plans to invest $150 billion in data center-related infrastructure over 15 years.
Professor Roberto Verdecia of the University of Florence in Italy mentioned, "There is a method to slightly sacrifice precision for energy efficiency." On the other hand, David Rolnick, a computer science professor at McGill University in Canada, warned, "We should not assume that big tech will solve the energy consumption issues related to large language models (LLMs) and then have society bear the consequences." Professor Rolnick cautioned, "We should not assume that companies will prioritize public interest over profit."
Meanwhile, Amazon Web Services (AWS) announced plans to build its first proprietary data center in Seo-gu, Incheon, South Korea, and invest about 8 trillion won in domestic cloud infrastructure by 2027.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
