NVIDIA (NASDAQ:NVDA) made a big move in data centers at its annual investor day in April by announcing the Grace CPU (central processing unit) — a product that could add more fuel to the chipmaker’s rapidly growing data center business.
Grace is NVIDIA’s first data center CPU, which would mark its entry into a market dominated by Intel (NASDAQ:INTC) when it goes on sale in 2023. The company claims Grace can “deliver 10x the performance of today’s fastest servers on the most complex AI and high-performance computing workloads.” But the more important thing to note here is that it would add another dimension to NVIDIA’s data center chip lineup, making it one of the best ways to play a fast-growing industry. Let’s see why.
NVIDIA targets the data center accelerator market
Data center accelerators such as GPUs (graphics processing units), CPUs, ASICs (application-specific integrated circuits), FPGAs (field-programmable gate arrays), and DPUs (data processing units) are in massive demand.
According to a third-party report, the data center accelerator market could clock a compound annual growth rate (CAGR) of nearly 47% through 2027. The market can generate nearly $98 billion in annual revenue at the end of the forecast period, driven by the growth of high-performance computing (HPC) and artificial intelligence workloads in the cloud.
NVIDIA has already been taking advantage of this market with its data center GPUs and recently stepped up its game in the market with a new offering in the form of data processing units (DPUs). Its chips have been in great demand thanks to their ability to accelerate intensive workloads such as AI, deep learning, and big data analytics. Industry website The Next Platform estimates that GPUs account for 90% to 95% of the overall computational power of the servers they are installed in, which explains the 124% growth in NVIDIA’s data center revenue last fiscal year to $6.7 billion.
With the launch of its server CPUs, NVIDIA will be able to dive into a market that could hit $19 billion in revenue by 2023, according to rival Advanced Micro Devices‘ (NASDAQ:AMD) estimates. In fact, the server CPU opportunity is supposed to be substantially bigger than the data center GPU market that’s expected to produce only $11 billion in revenue by 2023.
So, NVIDIA will substantially boost its addressable market with the launch of the Grace CPU. But will it be enough to claw away market share from established players such as Intel and AMD?
Why AMD and Intel need to be worried
Intel is the dominant player in the CPU market with a share of over 90%. However, it has been losing share to AMD since the launch of the EPYC processors in 2017. AMD has made yearly improvements to its EPYC server processors over the years, and they are now in their third generation.
The recently launched EPYC 7003 processors have been a hit with consumers. AMD forecasts that its server platform could end 2021 with 400 cloud instances (a virtual server in a cloud computing environment) as compared to 200 last year. The chipmaker also anticipates 100 new server original equipment manufacturers (OEMs) to deploy EPYC processors this year.
Not surprisingly, AMD has been consistently taking away server market share from Intel over the years as the latter fell behind the technology curve. Chipzilla is looking to catch up to AMD with its recently launched 10-nanometer (nm) Ice Lake SP data center CPUs. Intel will look to further reduce the technological gap with AMD by moving to the Sapphire Rapids architecture in the next two years.
NVIDIA, however, can turn this two-horse race into a three-party struggle. That’s because NVIDIA already sells DGX and HGX server systems that are powered by multiple GPUs and utilize AMD or Intel’s chips, giving it an existing ecosystem where it can showcase Grace. NVIDIA claims that using the Grace CPU instead of those from its rivals in its server systems would translate into 30 times higher aggregate bandwidth, and a tenfold jump in performance over the current DGX systems.
So, NVIDIA already has the platform to potentially upstage its rivals once Grace hits the market. If the chipmaker can deliver the performance jumps it is claiming, it won’t be surprising to see it take market share away from Intel and AMD.
All of this should bode well for NVIDIA’s data center business, which supplied 40% of its total revenue last year. The segment now has an additional growth engine that could help the chipmaker remain a top tech stock for a long time to come.
This article represents the opinion of the writer, who may disagree with the “official” recommendation position of a Motley Fool premium advisory service. We’re motley! Questioning an investing thesis — even one of our own — helps us all think critically about investing and make decisions that help us become smarter, happier, and richer.
View more information: https://www.fool.com/investing/2021/04/28/why-nvidia-data-center-move-should-give-amd-intel/