Arm Aims to Capture 50% of Data Center CPU Market by 2025, Riding the AI Wave

Arm aims to capture 50% of data center CPU market in 2025

Arm Holdings hopes to increase its share of the global data center CPU market from 15% to 50% by the end of 2025. Mohamed Awad, senior vice president of infrastructure at Arm, made the claim in an interview with Reuters. The company pins its hopes primarily on AI servers, so consider offerings like Nvidia's GB200 and GB300 machines, custom silicon from large cloud service providers, and Ampere Computing-based systems. Most servers today run AMD's EPYC processors or Intel's Xeon CPUs that rely on the x86 instruction set architecture, as there is more data center-grade software for x86. However, the situation is changing, and Arm says that some server programs are now developed for Arm-based processors first and then ported to x86. Google and Microsoft have also started designing data center processors with Arm's technology, although their projects are at an earlier stage compared to Amazon. While x86 dominates the server market and will likely continue for a while, Arm adoption is growing. Arm is used by Amazon Web Services for its Graviton CPUs, which are used for many of its instances instead of AMD's or Intel's processors. Half of the processors used by AWS are indeed Arm-based Graviton CPUs. In addition, Ampere Computing offers Arm-based CPUs for data centers. Ampere is a chip designer backed initially by Oracle but now owned by SoftBank (which also happens to own Arm Holdings). In addition to AWS and Ampere, Nvidia is emerging as a major backer of Arm in the datacenter space. The company's Grace CPUs with 144 Arm Neoverse V2 cores power GB200 and GB300 AI servers will likely become popular with large cloud service providers. But Arm pins its hopes not only on AWS, Ampere, and Nvidia. The company also offers compute subsystems (CSS) based on its Neoverse cores, enabling chipmakers to build their data center-grade CPUs relatively easily. Furthermore, Arm is reportedly developing its own CPUs for large cloud service providers, such as Meta. These CPUs have yet to gain market share, though if Meta deploys them in volume, they will inevitably control a significant part of the server CPU market as Meta is one of the major users of servers globally.

Arm targets 50 percent of datacenter CPUs this year

Arm expects to see its architecture account for half of the datacenter CPU market by the end of this year, up from 15 percent in 2024, all thanks to the AI boom. The Brit chip designer has long touted the AI credentials of chips based on its architecture and is gradually gaining ground in the server market, but Arm's infrastructure chief, Mohamed Awad, now claims that he expects its share of the global market for datacenter processors to surge to 50 percent by the end of the year. In an interview with Reuters, Awad claimed that Arm's technology typically offers lower power consumption than processors made by rivals such as Intel and AMD. As a result, it was argued, Arm-based products have become increasingly popular even with the big cloud computing companies, which are concerned about the power drain of their massive bit barns. But going from 15 percent last year to 50 percent by the end of this year would be a massive leap. We asked Arm to confirm if Awad did actually claim this, and how it expects to reach this figure. The Softbank-owned biz said it is largely based on the growth of AI servers. "In this unprecedented age of AI transformation we are seeing an insatiable demand for computing, with AI servers set to grow by more than 300 percent in the next few years," Awad said in a follow-up note to The Register. "For that to scale, power efficiency is no longer a competitive advantage - it is a baseline industry requirement. This is where the Arm Neoverse compute platform is the clear leader, and the compute platform of choice for industry-leading partners including AWS, Google, Microsoft, and Nvidia," he claimed. The chip designer has been seeing its architecture increasingly deployed by the big three cloud operators just mentioned. In 2023, Bernstein Research estimated that nearly 10 percent of servers across the world contain Arm application processors as their primary brains, and half of those were deployed by Amazon, which said it had more than two million of its custom Graviton chips in the cloud. Microsoft and Google were relatively late to the custom Arm processor party, with the Chocolate Factory basing its Axion chip on the designer's Neoverse V2 blueprint, while Redmond announced general availability of its own Cobalt 100 Arm CPUs in the Azure cloud later in the year. Scaling deployments by all three cloud providers could account for some of the increase that Awad is forecasting for this year, but Nvidia products are also likely to account for a significant share. The DGX GB200 NVL72 rack system comprises 36 of Nvidia's Grace CPUs and 72 Blackwell GPUs, for example, making for 2,592 Arm Neoverse V2 cores in each unit deployed, and these are likely to be in demand this year. But it is also easy to overlook other products in the datacenter that have Arm-based cores inside them, such as SmartNICs and DPUs (data processing units) such as Nvidia's BlueField-3, and also the Nitro server management cards in AWS servers. Arm's AI ambitions also extend beyond the datacenter, as chief exec Rene Haas enthused in an earnings conference call late last year. "I think when we look at what's going on with AI and when you think about AI, it's not just training in the datacenter, but it's inference in the datacenter, it's inference across different parts of the overall value chain, the network, the automobile, the PC, the mobile phone, the wearable, which can be kind of what people would call the edge," he claimed. Arm's parent SoftBank recently announced its intention to acquire Ampere, a chip firm that sells server processors based on the Arm architecture. Ampere told The Register it is expanding out from the cloud companies it initially targeted with an eye on the telecoms market. ®

Exclusive: Arm expects its share of data center CPU market sales to rocket to 50% this year

SAN FRANCISCO, March 31 (Reuters) - Arm Holdings (O9Ty.F), opens new tab expects its share of the global market for data center central processing units to surge to 50% by the end of the year, up from about 15% in 2024 with gains driven by the boom in artificial intelligence, a senior executive said. Arm's CPUs are often used as a "host" chip inside of an AI computing system and act as a kind of traffic controller for other AI chips. Nvidia (NVDA.O), opens new tab, for example, uses an Arm-based chip called Grace in some of its advanced AI systems which contain two of its Blackwell chips. Arm's tech in many cases offers lower power consumption than rival processors made by Intel (INTC.O), opens new tab and Advanced Micro Devices (AMD.O), opens new tab, Mohamed Awad, Arm's infrastructure chief, told Reuters. As AI data centers use huge amounts of electricity, Arm's chips have become increasingly popular among cloud computing companies. Awad added that data center chips often use more of Arm's intellectual property and the company typically receives "a lot higher aggregate royalty rate" than chips for less complex devices. UK-headquartered Arm, which is 90% owned by Japan's SoftBank Group (9984.T), opens new tab, does not make chips itself but sells the fundamental building blocks and other intellectual property to cloud computing companies and firms like Apple (AAPL.O), opens new tab and Nvidia to use to design chips for laptops, smartphones and data center processors. Arm generates revenue by billing companies for a license to use its tech and collects royalty payments for each chip sold. It struggled to make headway in the lucrative data center market for nearly two decades as switching over from the once-dominant x86 chips made by Intel and AMD means clients have to rewrite software as well as change up parts of hardware. "We've gotten to the point where software is actually being developed for Arm first and foremost," said Awad. Amazon.com (AMZN.O), opens new tab has designed in-house data center CPUs with Arm tech that accounted for more than half of the capacity for chips it added over the last two years, Amazon said in December. Alphabet's (GOOGL.O), opens new tab Google and Microsoft (MSFT.O), opens new tab have also made Arm-based data center chips, though their efforts are more recent than Amazon's. Reporting by Max A. Cherney; Editing by Edwina Gibbs Our Standards: The Thomson Reuters Trust Principles., opens new tab Suggested Topics:Artificial Intelligence Max A. Cherney Thomson Reuters Max A. Cherney is a correspondent for Reuters based in San Francisco, where he reports on the semiconductor industry and artificial intelligence. He joined Reuters in 2023 and has previously worked for Barron's magazine and its sister publication, MarketWatch. Cherney graduated from Trent University with a degree in history.

ARM targets 50% of the AI market in 2025

British chipmaker ARM aims to triple its share of the artificial intelligence data center market by 2025. This ambition is underpinned by its original business model, growing compatibility and attractive energy efficiency. ARM Holdings, a company that is 90%-owned by SoftBank, specializing in chip architectures, is planning a spectacular breakthrough in the artificial intelligence sector. According to its latest projections, the company aims to increase its market share in AI data centers from 15% to 50% by next year. This is indeed a bold ambition, although is one that builds on a momentum that is already well underway. Royalties, design and efficiency Unlike manufacturers like Intel or Nvidia, ARM does not produce chips. Its model is based on the sale of designs and technical elements, accompanied by a system of royalties. These royalties are particularly lucrative in the field of AI, being higher than in traditional systems. As a result, giants such as Nvidia or Alphabet can offer chips stamped with their own name, while relying on ARM architecture. This is notably the case for the Grace chip, used in several Nvidia systems. At the same time, the energy efficiency of ARM architecture represents an increasingly valuable asset, as the energy cost of AI becomes a major issue in industry. For a long time, ARM's development was restrained by the dominance of x86 architecture, used by Intel and AMD. This ubiquitous standard made it difficult to switch to other architectures. However, the situation changed with Apple 's transition to its ARM-based M1 chips, followed by growing adoption in data centers. "We've reached the point where software is developed for Arm first and foremost," says Mohamed Awad, head of infrastructure at ARM. The movement is supported by several giants: Amazon has designed ARM chips in-house that represent over half of the capacity added to its data centers over the past two years. Google and Microsoft are following suit with more recent projects. With a lean business model, an increasingly adopted technology and strong partners, ARM seems well positioned to change dimensions in AI. It remains to be seen whether the pace can keep up to expectations.