Counterpoint Research noted on Thursday that the AI infrastructure market is undergoing a structural shift as hyperscalers increasingly move away from Intel Corp. (NASDAQ:INTC) and Advanced Micro Devices Inc.’s (NASDAQ:AMD) legacy x86 central processing units (CPUs) toward proprietary Arm Holdings plc (NASDAQ:ARM) Arm-based designs to optimize cost, efficiency, and control.

Hyperscalers Drive Shift From x86 to Arm

Major cloud players such as Alphabet Inc. (NASDAQ:GOOGL) Google, Amazon.com Inc. (NASDAQ:AMZN) Amazon Web Services, Microsoft Corp. (NASDAQ:MSFT), and Meta Platforms Inc. (NASDAQ:META) are leading the transition by redesigning their AI server stacks around Arm-based CPUs, as per Counterpoint.

Historically, these companies relied on x86 processors from Intel and AMD due to software compatibility and existing infrastructure. However, the rise of custom AI accelerators has increased the need for heterogeneous architectures, accelerating the adoption of Arm-based CPUs built on Neoverse cores, the firm noted.

This shift reflects a broader strategy: hyperscalers are designing in-house silicon to reduce reliance on external vendors, improve margins, and lower the cost of running AI workloads at scale.

Arm Gains Ground With Efficiency and Custom Silicon

Arm Holdings is gaining traction as its architecture delivers significantly better performance per watt than traditional x86 systems, a critical advantage in power-constrained data centers, according to Counterpoint.

Companies are already deploying Arm-based CPUs across AI infrastructure. Google is scaling its Axion CPU for next-generation Tensor Processing Unit (TPU) systems, while AWS is expanding the use of its Graviton processors alongside Trainium chips. Microsoft has integrated its Azure Cobalt Arm CPU with its Maia AI accelerators, embedding Arm into its AI stack from the outset, the firm added.

These deployments show that Arm is no longer limited to general-purpose cloud workloads but is becoming central to AI server design.

Custom AI Infrastructure Reshapes Competitive Landscape

The transition is unfolding generation by generation, with hyperscalers aligning CPU design closely with their proprietary AI accelerators. Meta Platforms has reinforced this trend by selecting Arm as a strategic partner for its next-generation Meta Training and Inference Accelerator (MTIA) infrastructure and serving as the launch customer for Arm’s Artificial General Intelligence (AGI) CPU platform, Counterpoint told.

This coordinated shift is expected to accelerate from the second half of 2026, driven by the broader deployment of in-house Arm CPUs. Projections indicate Arm-based CPUs could account for around 90% of host CPU deployments in custom AI Application-Specific Integrated Circuit (ASIC) servers by 2029, up from about 25% in 2025, the firm told.

As hyperscalers scale their in-house silicon strategies, the impact is extending across the semiconductor supply chain, with rising demand for advanced manufacturing supporting both AI accelerators and Arm-based CPUs.