AMD and ARM Overtaking Intel in Server CPU Market, Says DIGITIMES Research

In recent years, AMD and ARM have been making significant strides in the server CPU market, outpacing the long-time leader, Intel, according to Frank Kung, a DIGITIMES Research analyst. One of the main factors behind the shift would be the price of AMD’s products, which can be at least 30% lower than Intel’s for comparable hardware. AMD’s CPUs also have a higher number of cores, making them ideal for server environments. Meanwhile, the growth of ARM-based CPUs in the server market has been slower, but still aggressively pursued by cloud owners and server manufacturers.

According to Frank Kung, a DIGITIMES Research analyst specializing in the server industry, AMD and Arm have been overtaking Intel in the server CPU market in recent years, and the margins of the share that AMD had won over were especially large in 2022 as data center operators and server brands started finding that solutions from the number-2 maker were getting better than those of the long-time leader. Frank Kung projects that AMD’s share will well stand above 20% in two years.

One of the three main factors that led data center owners and server manufacturers to move to AMD was price. According to Kung, when comparing server processors from AMD and Intel with comparable core counts, clock speeds, and hardware specs, the prices of most of the former’s products are at least 30% lower than those of the latter, and the price disparities may exceed 40%.

As server businesses typically buy their processors in big quantities, this disparity would significantly lower their costs if they chose AMD’s options. According to Kung, server businesses need not be concerned about interoperability because both Intel and AMD’s CPUs are built on the x86 design.

Arm-based CPUs

The large number of cores found in AMD CPUs also makes them ideal for server environments because a CPU’s handling capacity increases as the number of cores increases. The first of AMD’s 128-core EPYC processors is scheduled to appear in the first half of 2023, while Intel’s finest option in terms of core count currently stands at 60. AMD’s 96-core EPYC processor, which was designed using Genoa architecture, was released in the fourth quarter of 2022.

Second driving force is TSMC support. According to Kung, all of AMD’s server CPUs are produced by TSMC using the most recent manufacturing techniques, which allows them to deliver top-notch performances. Kung also stated that because of TSMC’s cutting-edge technologies and high yield rate, AMD has never had a problem with missing its product launch schedule. With Intel, this is not the situation.

– story continues below the image –

The reality that Intel manufactures all of its premium processors on-site serves as the third motivator. Intel’s internal manufacturing technologies have been rather unreliable over the past few years, according to data from Intel’s source vendors, while server manufacturers and datacenter operators have frequently observed Intel postponing the volume production timetable of its new server platform.

Microsoft and Google are the two cloud owners who are most interested in purchasing computers running AMD’s products. The two cloud service companies currently buy over 30% of AMD-based server versions, and HP Enterprise (HPE) is more interested in AMD-powered servers than any other server brand.

According to Kung, the market share growth of Arm-based CPUs in the server market was a little weaker in 2022 than that of AMD-based ones, and the growth will decline even more in 2023. However, in the long run, there would still be significant development potential for Arm-based CPUs.

Although x86-based Processors from AMD and Intel can compete with Arm-based ones in terms of processing speed while using much less power, interoperability is presently their greatest flaw.

Since the majority of server programs are built using the x86 architecture, the issue is unlikely to be resolved until more servers that are based on the Arm architecture start to appear, luring more middleware developers to the market and encouraging them to create solutions to translate x86 codes for Arm systems.

The growth of Arm CPUs in the server industry, however, is still being aggressively pursued by cloud owners and server manufacturers. Before 2022, Amazon and Alibaba began developing Arm-based goods, Microsoft and Google started working on Arm-based projects the following year, and HPE is increasing the use of Arm-based servers. The Arm design is currently being pushed by Nvidia and Ampere to be supported by their Arm-based processors. The demand for large-scale datacenter and edge computing systems will increase in the future years, which will boost the ESG potential for Arm Processors, Kung noted.

About the Analyst

Frank Kung holds a bachelor’s degree in electrical engineering as well as an MA in Technology Innovation Management. He has more than ten years of expertise conducting research and managing projects at organizations such as the Institute of the Information Industry and industrial Computer companies. His current study focuses on the server business and application market patterns, such as those in HPC, AI, data centers, and peripheral computing.

About DIGITIMES

Since 1998, the media and research company DIGITIMES has been offering insightful information and analysis to a large audience of investors, researchers, and high-tech industry leaders. For those who need to remain informed about the most recent advancements in the business, DIGITIMES is a reliable source of information. For clients in the industry’s upstream, downstream, and end markets, DIGITIMES Research provides research data, output and sales data, expert analysis, as well as advisory services to help them understand macrotrends and future-oriented value.