WHILE AIR COOLING STILL HAS A ROLE TO PLAY , IT OFTEN FALLS SHORT IN COOLING AI HARDWARE EFFICIENTLY . if not outright economically impossible , leading to huge problems down the line .
FEATURE
What are the key areas that AI will change in data centre design ?
AI has transformed data centre design requirements . The adoption of AI applications is pushing data centres to evolve , requiring specialised infrastructure that caters to the unique demands of Machine Learning algorithms . Three key areas will need to evolve to keep up with increasing worldwide AI spending .
First and foremost , there ’ s a huge increase in the appetite for power . AI computation relies heavily on specialised processors , particularly GPUs ( Graphics Processing Units ) and DPUs ( Data Processing Units ), which draw considerably more power , emit more heat and occupy more space compared to traditional CPU-powered racks .
Secondly , this means data centres must invest in robust power infrastructure and devise efficient cooling solutions to prevent overheating to combat relentless energy consumption . While air cooling still has a role to play , it often falls short in cooling AI hardware efficiently . Often , liquid cooling is the preferred option for high-performance chips , offering superior cooling and potential cost benefits .
Finally , cabling – often the ‘ forgotten child ’ of design – is also habitually overlooked and under-budgeted . As with power connections and cooling systems , cabling is embedded infrastructure . These are all built into the structure of a data centre complex . This means systems can be extortionately expensive to replace ,
WHILE AIR COOLING STILL HAS A ROLE TO PLAY , IT OFTEN FALLS SHORT IN COOLING AI HARDWARE EFFICIENTLY . if not outright economically impossible , leading to huge problems down the line .
What are the consequences of cutting corners in data centre design ?
With new business technologies – particularly AI – emerging at a rapid pace , data centres are facing an unprecedented surge in demand . In response , many operators may be tempted into shortcuts as a way of reducing costs . While choices like these may provide immediate savings , they unknowingly plant the seeds for significant long-term costs and problems , such as outages or downtime .
When future-proofing data centres , designers cannot afford to look at the individual pieces of the puzzle – from power to cabling – in isolation . Design decisions can have a lasting impact on the business landscape .
Once data centres are operational , changing these design aspects is costly and potentially irreversible . As a result , cutting design corners can lead to enduring issues like frequent downtime and limited scalability , and recognising this can help operators plan for future design needs and the required infrastructure .
What areas of data centre design are often overlooked ?
Historically , data centre design has revolved around power and cooling requirements , while areas such as cabling
38 www . intelligentdatacentres . com