The world’s data centres are on the verge of a decade long upgrade cycle as they retool and adopt high performance computing. Who’s going to do most of the work? Probably generative AI.
That’s according to NVIDIA CEO, Jensen Huan.
“You’re seeing the beginning of call it a 10-year transition to basically recycle or reclaim the world's data centres and build it out as accelerated computing,” Mr Huan said on a recent earnings call. “You’ll have a pretty dramatic shift in the spend of the data centre from traditional computing, and to accelerated computing with smart NICs, smart switches, of course, GPUs, and – and the workload is going to be predominantly generative AI,” he added.
Citi research analysts say an acceleration of AI workloads and related IT infrastructure deployments are likely to be additive to the demand for data centre deployments for at least the first few years of this cycle, while the demand for power density could trend higher.
AMD recently hosted its Data Centre & AI day at which it sized the AI GPU data centre opportunity at $150bn by 2027. That’s up from $30bn in 2023.
Citi research analysts said they don’t expect existing general computing workloads to initially be cannibalized by the accelerated IT infrastructure that is more likely to be focused- at least initially- on an expanding range of learning and inference Generative AI models.
Over the longer-term, they added, there may be some degree of data centre resource optimization to allow for a potentially greater mix of accelerated computing relative to general-purpose computing.
Citi research analysts also said the acceleration of high-performance computing adoption could be a potential positive in the mosaic of data centre demand, but nonetheless it’s a trend that is hard to quantify. They point out that this is not dissimilar to how other technology themes have influenced data centre demand,
For example, digital transformation and adoption of hybrid-cloud architectures by enterprise customers are helping demand for the broader data centre ecosystem. But they add that from top-down perspective, these trends are difficult to quantify including for more on that see the full report.
For more information on this subject, please see US Telecommunications Infrastructure - AI Implications for Data Centre Stocks (31 May)
Citi Global Insights (CGI) is Citi’s premier non-independent thought leadership curation. It is not investment research; however, it may contain thematic content previously expressed in an Independent Research report. For the full CGI disclosure, click here.
