AI Reshapes Energy Efficiency in Networking and Data Centers

The article asserts that AI both drives rising energy demand and provides tools to improve energy efficiency across networking and data centers.

• Main claim: AI is simultaneously the primary driver of rising data-center electricity demand and a powerful efficiency tool; the IEA projects global data center electricity consumption will nearly double to ~945 TWh by 2030 (roughly equivalent to Japan’s annual electricity use). The piece cites Deloitte research showing GPU power per chip grew from ~400 W in 2022 to >1,200 W for advanced processors, and references an SD-WAN market projection of $35 billion by 2030.
• Supporting details and concrete technologies: AI-driven approaches highlighted include AI-assisted telemetry, SD-WAN traffic prioritization, predictive maintenance, network slicing, and machine-learning cooling optimization; specific infrastructure shifts noted are direct liquid cooling, denser rack configurations, and reassessment of optical vs copper interconnects. The article frames modern colocation and software-level techniques like model distillation as practical responses, with timelines anchored by the 2030 market and IEA projections.