Will the desire for artificial intelligence technology take precedence over the health of the planet? You can’t have both at the moment. No matter how many solar panels and electric vehicle chargers a building has, its outstanding sustainability record cannot offset the construction of a large-scale data center.
AI revolution and its requirements
The rise of AI can be said to be truly explosive. Companies are actively adopting generative AI to automate processes, gain insights from big data, and provide personalized customer experiences. These developments come at a cost. AI applications require massive processing power, increasing demand for data center capacity. Complex AI models consume a lot of energy. This type of consumption runs counter to the sustainability narrative the cloud industry once touted. Today, enterprises and cloud service providers are actively avoiding conversations about sustainability.
Don’t get me wrong. I think some companies are still doing a good job of curbing their carbon footprint and should be celebrated. What I am talking about is generality across the industry. Because the larger discussion about sustainability suddenly faded into the background. The media also rarely asks questions. Most companies have removed sustainability from their promotional programs.
Rush to build data centers
The need for rapid expansion created a paradox. As cloud data centers grow in footprint and processing power, providers are moving further away from the sustainability ideals they once espoused. Many companies are in difficult situations. AI capabilities must be leveraged to remain competitive, but these realities cannot be reconciled with commitments to sustainability.
On the one hand, companies want to demonstrate progress in sustainability and even increases in environmental, social, and governance (ESG) scores. Although many businesses still view AI as an option, AI will be introduced to every business in some form, and there is no way to avoid the amount of power it will consume.
You can tell just by looking at the difference in power consumption between GPU and CPU. The CPU was suitable for most traditional processing, both in the cloud and on-premises. However, the shift to AI has made GPUs the preferred processor. GPUs consume more power because they are designed for the parallel processing required for graphics rendering and intensive computation in AI and games.
High-end GPUs use 200 to 450 watts or more, while high-end desktop CPUs use 65 to 150 watts. Mid-range GPUs use about 100 to 250 watts, while mid-range CPUs typically consume about 65 watts. Replacing CPUs with GPUs requires significantly more power, cooling, and data centers. Companies are reporting high demand for AI. Doing the math quickly shows that there will not be enough electricity produced on Earth to meet these future demands.
sound of silence
Discussions about the environmental impact of massive data center expansion can undermine the appeal and seemingly limitless potential of AI. Therefore, many companies are conducting public discussions focusing on the innovative and transformative potential of AI rather than environmental issues.
The rapid pace of technological advancement will outpace the advancement of sustainable practices and technologies. Current renewable energy sources and energy efficiency improvement measures are needed to help keep pace with the exponential growth in data processing demand. Given the speed and investment in AI, its growth will soon become unsustainable.
What is of concern at this time is how quickly the transition away from sustainability has occurred. It seems like just yesterday that I was presenting a sustainable cloud computing strategy at a conference. There was even a conference devoted solely to sustainability. The industry must rethink the sustainability of cloud computing and traditional enterprise data centers.
Successfully addressing the environmental impact of AI requires innovative solutions that balance technical capabilities and environmental responsibility. This includes more efficient chip designs, advances in cooling technology, and increased investment in sustainable energy sources to effectively power data centers.
Just because we stop discussing sustainability does not mean that sustainability issues have disappeared. Let’s put aside for a moment the worry that artificial intelligence-led cyborgs will take over the world. The scary news about AI right now is the real threat to the power grid. Now is the time to restart the discussion about sustainability.
editor@itworld.co.kr
Source: www.itworld.co.kr
