Hyperscale data-center capacity on pace to triple over next six years

Uncategorized

The rush to embrace artificial intelligence, particularly generative AI, is going to drive hyperscale data-center companies like Google and Amazon to nearly triple their capacity over the next six years.That’s the conclusion from Synergy Research Group, which follows the data center market. In a new report, Synergy notes that while there are lots of exaggerated claims around AI, there is no doubt that generative AI is having a significant impact on IT markets.Synergy’s updated six-year projection shows that the typical capacity of new hyperscale data centers will quickly be more than double that of existing operational hyperscale information centers. And the overall capacity of all operational hyperscale data centers will grow practically threefold between 2023 and 2028. The research study is based upon an analysis of the data center footprint and operations of 19 of the world’s major cloud and web service companies. By mid-2023, those business had 926 major information centers in operation around the globe. Synergy’s forecast for future data centers consists of another 427 facilities in the coming years.Synergy states the effect of generative AI advances has not so much sustained an increase in the number of information centers however rather resulted in a significant boost in the amount of power required to run those information centers. As the number of GPUs in hyperscale data centers skyrockets, driven primarily by AI, the power density of associated racks and data center facilities also requires to increase substantially. This is triggering hyperscale operators to rethink some of their data center architecture and implementation plans.So if it’s a headache for AWS to handle this things, what will it be like for a typical enterprise running five-year-old servers? While corporations are rushing to welcome generative AI to improve their businesses, the cost of obtaining the hardware and operating it has actually given many pause. A DGX server from Nvidia, custom-built for generative AI and loaded with hardware, can quickly run in the six-figure variety. For that sort of cash, you can likewise buy about 10 regular servers. Which will business prioritize?Plus there’s the expenditure of operating them. Nvidia GPUs are not known

for being a low power draw. It’s quite the opposite. They are the biggest power hogs in a data center. So for a budget-conscious business, especially a midsized one, deploying generative AI hardware might be too demanding. In addition, the way AI runs is different from conventional line-of-business applications. There’s the process-intensive task of training, which requires GPUs, and then there’s inference, which runs off the models trained by the GPUs. When a model is trained, there’s a likelihood you may not require to modify it for months. Then your extremely expensive hardware sits idle, unnecessary, and depreciating.Could business do this themselves without utilizing a hyperscale cloud service provider?”In theory, yes, but expenses may be prohibitive and access to the right sort of proficiency might be seriously restricted,”said John Dinsdale, chief analyst and research director at Synergy Research Group.So the emerging pattern in business IT for generative AI is to farm out the training portion of its AI however do the inference, which is much less

process intensive, internal. Why invest hundreds of countless dollars in hardware that you only utilize sparingly when you can rent it out from Google or AWS?This is known as AI as a service, an emerging … Source

Leave a Reply

Your email address will not be published. Required fields are marked *