Samil PwC "Electricity Consumption Due to AI to Double in 4 Years"

'AI Eats Electricity to Grow' Report
Surge in Power Demand Expected with AI Technology Advancement
Energy Savings in Other Fields through AI Use as a Solution

Samil PwC announced on the 24th that it has published a report titled "AI Grows by Consuming Electricity: Causes and Implications of Increased Power Consumption in the AI Era," which reviews the current status of power demand and suggests forecasts and implications in the era of generative artificial intelligence (AI). Detailed contents of the report can be found on the Samil PwC website.

According to the report, as AI technology leads innovation across a wide range of fields, power demand is expected to surge in the future. This is because large-scale data processing and high-performance computing capabilities are required to advance AI models, resulting in significant power consumption during these processes.

Since AI systems are primarily installed and operated within data centers, the expansion of AI ultimately leads to increased power consumption in data centers. According to the International Energy Agency (IEA), global data center power usage is projected to more than double from 460 terawatt-hours (TWh) in 2022 to 1,050 TWh by 2026.

The report identifies learning and inference capabilities as the reasons why AI consumes a large amount of electricity. Complex computations, deep learning model advancements, and large-scale datasets are required during learning and inference, and real-time processing demanded in the inference process also consumes vast power resources. In particular, the report adds that the power demand for cooling to dissipate heat generated by performing multiple computations simultaneously further drives power consumption.

As a response strategy, the report suggests adopting the principle of 유취단장 (有短取長), meaning "accepting the shortcoming but taking the advantage," by not overly worrying about the drawbacks but actively embracing the benefits. The first approach is to reduce the power usage of AI itself, and the second is to use AI to save more energy in other fields. Smart buildings, which always optimally configure building air conditioning systems and lighting, are a representative example. Regarding this, the report recommends, "Rather than worrying about AI's own power consumption, efforts should be made to develop technologies that use AI to save energy in other sectors."

This content was produced with the assistance of AI translation services.