L A T E S T I N T E L L I G E N C E

PRESENTED BY

INTRODUCTION

The data center landscape is undergoing a seismic shift due to the growth of artificial intelligence . Gartner notes that 70 % of executives report exploring generative AI for their organization ; IDC forecasts that the AI software market is expected to be worth nearly $ 792 billion by 2025 and is projected to grow at a rate of 18.4 % through that year .

This growing adoption leads to profound changes in data center power requirements and infrastructure planning . As AI technologies advance , data centers ’ power demands and infrastructure requirements must evolve to accommodate increasing power densities , changing hardware configurations , and dynamic workloads .

Data centers must be agile enough to scale up or down quickly to meet fluctuating demands without disrupting operations . This adaptability ensures efficient resource utilization , future-proofs the facility , and creates a scalable data center for seamless expansion to meet growing computational needs . Together , adaptability and scalability are essential cornerstones in designing data centers that can effectively harness the potential of AI while staying responsive to evolving business and technology trends .

DESIGN CRITERIA

The average data center design RFP in the late 2010 ’ s required approximately 5kW per cabinet and grew to 10kW per cabinet from approximately 2012 to 2018 . As an industry we quickly moved on from single phase 120-volt inputs , spent several years at single phase 208 volts , moved into three phase 208 volts , and on to other voltages , like 415 volts . The voltage was coming up , and the current , protective devices , and conductor sizes were all going down . Select applications in the traditional data center environment started ticking up , and per rack power would start to increase to 10 – 20 % of the overall environment . Designs began to bifurcate into density and availability zones where we could leverage power distribution and heat rejection technologies based on the various power densities required in an overall space . All the while utilization was lagging design , presenting an inefficient balance on space utilization , and mechanical and electrical systems . �

20 www . intelligentdatacentres . com