Hong Kong data centres face rising heat and power demands from AI
Artificial intelligence is driving increased power consumption and heat production in Hong Kong data centres, complicating cooling efforts as the city experiences rapid warming. Analysts suggest that operators may need to move beyond mandatory energy efficiency requirements to establish absolute energy, water, and carbon targets.
Starting September 20, the Hong Kong government will extend mandatory energy efficiency requirements to data centres, necessitating energy audits every five years and the disclosure of technical information. However, current regulations do not require operators to implement specific measures identified in audits, nor is there a set limit on power efficiency.
The challenge is compounded by rising local temperatures; the Hong Kong Observatory reports the city has warmed by 0.35°C per decade since 1996, more than double the rate seen in previous long-term periods. Because AI systems can consume up to eight times more power per server rack than older equipment, traditional air cooling is becoming less effective. Consequently, operators are increasingly shifting toward liquid cooling technologies to manage heat closer to processors.
Entities
Harry Lai · Hong Kong Green Building Council · Hong Kong Observatory · Jones Lang LaSalle Ltd. · Sandy Ridge