Arm's stock price soars 15% after launching AI chips

Arm Holdings (ARM) stock surged over 15% on Wednesday after the company officially announced its first mass-produced data center processor — the Arm AGI General Artificial Intelligence Central Processing Unit.

Traditionally, Arm licenses its intellectual property only to other companies for in-house chip development, with clients including Apple, Nvidia (NVDA), and others; Nvidia’s Grace and Vera CPUs are both based on Arm architecture.

Graphics Processing Units (GPUs) have long dominated the data center market due to their ability to train and run AI models. However, as AI inference scenarios become more common than training, and as the industry shifts toward intelligent agent applications, the importance of Central Processing Units (CPUs) is increasingly evident.

This creates an opportunity for Arm to launch its own processors. The company not only released the chip but also simultaneously introduced server racks capable of scaling up the operation of these chips.

Although Intel (INTC), AMD, and other vendors’ x86 architecture chips currently lead the data center market, Arm claims that its CPU single-rack performance can reach twice that of these platforms.

Arm Holdings CEO demonstrates the company’s AGI CPU.

Arm states that the AGI CPU is developed in collaboration with Meta, which will deploy this chip within its data centers, alongside its own custom chips.

In addition to Meta, Arm has partnered with Cerebras, Cloudflare (NET), F5 (FFIV), OpenAI (OPAI.PVT), Positron (POSC), Rebellions, SAP, and South Korea’s SK Telecom (SKM), among others, who will use the chip for AI and related intelligent agent applications.

Despite Wall Street’s enthusiastic response to Arm’s new chip, Bank of America Global Research analyst Vivek Arya pointed out in a report to investors that Arm is far from the only player in the CPU market.

He wrote, “It’s important to emphasize that the CPU market is highly competitive. Existing vendors with x86 and Arm architectures have more comprehensive product portfolios and mature software ecosystems capable of serving enterprise and telecom customers.”

Arya added, “Large-scale cloud providers all develop their own custom CPUs, and core customers like Meta and OpenAI have existing agreements with AMD and Nvidia for CPU collaborations. The market space for Arm AGI CPUs is limited. Moreover, as the AI industry grows larger, Arm’s smartphone and consumer electronics businesses face increasing memory supply pressures.”

Earlier this month, Meta and Nvidia announced an expansion of their partnership, with Nvidia providing the largest Grace CPU server deployment to date for the social media giant.

Just last week, AMD also announced a partnership with Meta, including server products equipped with the company’s Venice and next-generation Verano CPUs.

At Intel’s earnings call on January 22, CEO Pat Gelsinger highlighted AI as a primary driver of CPU demand.

Gelsinger stated, “The continued proliferation and diversification of AI workloads are creating significant capacity constraints for traditional and new hardware infrastructure, further emphasizing the increasingly important and indispensable role of CPUs in the AI era.”

At Nvidia’s GTC conference last week, CEO Jensen Huang focused on the upcoming Vera CPU, stating that the chip will be released alongside server racks to support AI applications.

The increased market attention on CPUs does not mean GPUs will be replaced. These high-performance processors remain essential hardware for running advanced AI models and will continue to play a key role in the foreseeable future.