Arm Holdings, long known as a low-power chip designer for mobile devices, is staking a bold claim in the high-stakes data center market. Its infrastructure chief, Mohamed Awad, predicts that Arm will seize 50% of the global data center CPU market by the end of 2025—up from just 15% in 2024. The catalyst? A rapid shift in AI workloads, energy-conscious cloud giants, and a maturing Arm software ecosystem that is finally ready for primetime.
The Rise of Arm in a x86 World
For years, Arm’s ambitions in the server market seemed idealistic. Intel’s x86 architecture held the keys to the kingdom, fortified by decades of software support and a hardware ecosystem that was nearly impossible to break. Arm-based CPUs, while efficient, lacked the horsepower and compatibility to compete in enterprise-grade environments. The cost of switching—rewriting software and retooling infrastructure—was too steep.
But AI changed the equation.
As AI workloads surged, so did power consumption. Traditional x86 chips began to look bloated and inefficient compared to Arm’s leaner, energy-saving designs. Cloud providers want efficiency, and Arm delivers that as a host CPU coordinating accelerators. Amazon, Microsoft, and Google are all on board, with AWS reporting that more than half of its new server CPUs are now Arm-based.
Breaking Through the Software Barrier
Arm’s real breakthrough has come not just from silicon, but software. The past few years have seen a transformation: new compilers, server-class designs, and tools that have made it easier to build for Arm. According to Awad, developers are now writing “Arm-first,” with many new applications optimized for the architecture from the ground up.
This shift is not merely technical—it’s strategic. By removing the friction to adopt Arm, cloud hyperscalers are now positioned to deploy it at scale. AWS’s Graviton series, Google’s Axion chips, and Microsoft’s Cobalt CPUs—all built on Arm—signal a new era where customization, efficiency, and AI-optimized performance matter more than legacy compatibility.
The Competitive Landscape: Arm, Intel, and AMD
Arm’s growth is coming at the expense of long-time titans.
Intel, whose Data Center and AI Group reported $3.4 billion in Q4 2024 (down 3% YoY), is facing the most acute pressure. Xeon server CPUs are losing ground, and efforts to reverse the trend with Sapphire Rapids and Sierra Forest chips have yet to regain momentum. Intel’s data center CPU sales recently hit a 13-year low.
AMD, meanwhile, is the strongest challenger. Its data center revenue surged to $12.6 billion in 2024, nearly doubling from the year prior, with its EPYC CPUs and Instinct AI GPUs gaining favor. With a 25% server CPU market share by late 2024, AMD has carved a meaningful foothold—but it’s also investing heavily in AI-specific hardware.
Arm, as an IP vendor, enjoys enviable margins. It licenses designs rather than manufacturing chips, earning royalties with every unit shipped. In fiscal Q3 FY2025, Arm posted record revenues of $983 million (up 19% YoY), with royalty revenues up 23%. Its Cloud & Networking segment, while still small (~11% of FY2023 revenue), is growing fast.
SoftBank’s Big AI Play
Much of Arm’s momentum can be traced back to its owner, SoftBank. After acquiring Arm in 2016 for $32 billion and seeing a failed $40 billion sale to NVIDIA, SoftBank took Arm public in 2023, retaining a 90% stake. Since then, it has doubled down on AI—investing in OpenAI, exploring AI hardware ventures, and using Arm as the cornerstone of its AI infrastructure strategy.
This long-term vision is paying off. With data center energy use expected to double by 2030 thanks to generative AI, and hyperscalers chasing every efficiency gain, Arm is uniquely positioned to thrive.
Reality Check: Can Arm Really Reach 50%?
Skepticism is warranted. Doubling market share in less than two years is rare. But the trajectory appears credible. Arm’s presence at Amazon, Google, and Microsoft gives it scale others can’t match. And as Deloitte and Snowflake have both pointed out, cloud providers are aggressively operationalizing AI, demanding energy-efficient, high-performance chips to manage escalating infrastructure costs.
If current deployment trends continue, especially among hyperscalers, Arm’s share of new server CPU shipments could very well approach the 50% mark by late 2025.\
A New Center of Gravity
Arm’s resurgence in the data center is more than a comeback—it’s a redefinition. From mobile roots to AI infrastructure linchpin, Arm has timed its pivot perfectly. Energy efficiency, AI readiness, and ecosystem maturity are aligning in its favor. For Intel and AMD, the battle is no longer just about speed or cores—it’s about relevance in a cloud-first, AI-driven world where every watt counts.
If Arm delivers on its promise, 2025 could mark the year it finally became the heart of the modern data center.
