NTT DATA: AI Expected To Consume More Than 50% Of Data Center Energy In 2028

Illustration of data centers (photo: Unsplash)

JAKARTA - The latest NTT DATA report entitled "Sustainable AI for a Greener Tomorrow", reveals that artificial intelligence (AI) technology requires enormous electrical power to train large language models (LLM), run inference processes, and maintain services that are always active.

Researchers estimate AI's workload will account for more than 50% of data center power consumption by 2028. Other environmental impacts include the use of water for data center cooling systems, electronic waste, and mining of rare minerals for hardware production.

NTT DATA's Head of Sustainability Innovation Headquarters, David Costa stated that although AI adoption is quite worrying, on the other hand, AI can also be used to minimize these risks.

"AI's extraordinary ability can help manage energy networks more efficiently, reduce emissions as a whole, predict environmental risks, and increase water conservation," David said in a statement.

Furthermore, a team of AI experts and consultants for the sustainability of NTT DATA called for the company not only to focus on AI performance metrics such as accuracy or speed, but also for sustainability purposes.

"AI that continues requires comprehensive thinking from start to finish, ranging from mining raw materials, production of hardware, implementation of the system, to final disposal," he added.

This sustainability responsibility must also involve various parties, ranging from hardware manufacturers, data center operators, software developers, cloud service providers, policy makers, investors, to consumers.

To overcome this, this report offers a number of the best practices, such as: