New research warns that the explosive growth of generative AI is increasing carbon dioxide emissions from data centers at an alarming rate. A recent report from Morgan Stanley predicts that by 2030, the global data center industry will be responsible for emitting 2.5 billion tons of greenhouse gases (GHGs).
This figure is three times higher than what would have been expected without the surge in demand for generative AI.
The rapid advancement of generative AI technologies, such as ChatGPT, has brought numerous benefits, but at a significant environmental cost.
With AI models requiring enormous computational power, the energy consumption of data centers has skyrocketed. As the world becomes increasingly reliant on cloud services and artificial intelligence, experts are concerned that these technologies could hinder global efforts to combat climate change.
Generative AI and Rising Data Center Emissions
In 2023, data centers emitted around 200 million tons of greenhouse gases. However, the Morgan Stanley report highlights that this figure could rise to 600 million tons by 2030 due to the growing demands of generative AI.
To meet this surge, more data centers are being constructed worldwide. These centers house servers that handle data for everything from streaming services to complex AI models.
The report emphasizes that 60% of these emissions come from the operations of data centers, which consume massive amounts of electricity. The other 40% is due to the manufacturing of construction materials and infrastructure used to build the centers.
Impact on Global Emission Targets
The tech industry, which includes data centers, currently contributes to 40% of the total annual emissions in the United States. This surge in emissions has brought the industry’s ability to meet net-zero targets into question.
Net-zero emissions mean balancing the amount of greenhouse gases emitted with the amount removed from the atmosphere, a goal many nations are striving to achieve to combat climate change.
Google, for instance, has already reported a 48% increase in its emissions over the last five years, primarily due to the rise of data-heavy technologies like AI. If the demand for AI continues to grow at this rate, achieving environmental goals may become even more challenging.
Cooling Systems and Water Scarcity
Data centers rely heavily on cooling systems to keep servers from overheating. One popular method is water cooling, where large amounts of water are used to reduce the heat generated by servers.
However, this solution comes with its own set of problems. Water is becoming an increasingly scarce resource in many regions, and using water cooling systems places additional strain on areas already facing high water scarcity.
This raises a dilemma for tech companies, which must balance their energy consumption with environmental sustainability. While water cooling helps reduce electricity use, it hampers green initiatives in regions facing droughts and water shortages.
Carbon Capture and Reforestation
Experts believe that advanced technologies like carbon capture, utilization, and sequestration (CCUS) could help mitigate the environmental impact of data centers. CCUS technology involves capturing carbon dioxide emissions and either using them in industrial processes or storing them underground to prevent them from entering the atmosphere.
However, this technology is still in its early stages and requires significant investment to become fully effective. According to Morgan Stanley, CCUS technology needs a $15 billion investment to meet global standards.
Reforestation projects are another potential solution. By planting trees, companies can offset their carbon emissions, as forests naturally absorb carbon dioxide from the air. However, while reforestation can help, it may not be enough to offset the massive emissions produced by AI-driven data centers.
