Intel and NVIDIA battle are locked in new battle for turf, the booming data centre market and at the heart of this skirmish the technology that’s changing the world: Artificial Interlligence (AI).
In the recent quarter ended April 30, NVIDIA’s revenue increased by 48% reaching $1.94 billion compared to previous year. A big revenue bump came from its Data centre business which recorded $409 million revenue in the first quarter of this fiscal, up 186% year-on-year.The reason for the exponential increase is the spike in demand for a specific kind of microprocessor called Graphic Processing Unit (GPU) made by NVIDIA.
Large technology companies like Google , Amazon, Microsoft , Facebook, IBM, and Alibaba have all installed NVIDIA’s elite Tesla GPUs to power their data centres to perform machine learning to analyse data gathered from the cloud and derive insights.
“We have seen the PC era, which was followed by the mobile era, and now we see the emergence of the AI era,” said Vishal Dhupar, MD, NVIDIA, adding, “Once viewed just as a gaming technology, GPUs are now making inroads into data centres driving initiatives around machine learning (ML) and artificial intelligence (AI)”.
Data centres are storehouses of large volumes of data which harness the compute power of GPUs to make sense of large repository of data. “The advent of GPUs have helped redefined the concept of systems design for many of the high-performance applications, both in the commercial as well as non-commercial segments of the market,” says Rajnish Arora, VP, Enterprise Computing Research, IDC.
Does this mean that Intel will soon be losing market share in the data centre space? Unlikely. “We do not believe that Intel is losing share of the server CPU market to NVIDIA… The numbers clearly demonstrate that Intel has consolidated its presence and market share in the server market,” Arora said.
Today, Intel CPUs power data centres all over the world and enjoys a near monopoly accounting for 95.7% of the total server shipped in 2016. Intel too is betting big on AI and has invested heavily in startups which help build AI and High-Performance Computing (HPC) capabilities into its forthcoming chips.
In 2015, Intel bought Altera for $16.7 billion, a company which makes programmable devices which can be optimised for AI and ML.
In August, it went ahead and bought deep learning startup Nervana Systems to bolster AI specific solutions.
Intel will start selling its new chips named `Knights Mill’ later this year to address the growing Machine Learning (ML) market.It’s also working on `Knights Crest’ chips which will handle deep neural networks, a component of deep learning (DL).
“Competition has always been around..” said Pradeep Dubey, Fellow at Intel Labs. “We are optimizing our chips for machine learning and deep learning applications as this is a huge opportunity for us. There are four new processors to be released this year”. These products will compete against NVIDIA GPUs in the data centre market.