Arm pushed the front lines into the heart of the x86 territory.

On September 8, at the Arm Everywhere China conference held in Shanghai, Arm updated its three product lines in one go: mobile, cloud, and physical AI. On the mobile side, it launched the second-generation computing subsystem CSS for Mobile 2; on the cloud side, it released Neoverse CSS N4 and continued to advance the AGI CPU released in March of this year; and on the physical AI side, it further expanded the Total Design ecosystem.

More noteworthy changes are occurring primarily in the cloud.

In the past, Arm primarily made money through architecture and IP licensing, leaving chip design to customers such as Qualcomm, Apple, Amazon, and Google. Now, from IP and CSS to AGI CPUs that can be delivered directly, Arm is gradually incorporating the processes that were previously left to customers into its own product range.

Some Chinese companies are also joining the Arm camp. Volcano Engine plans to introduce Arm's AGI CPU into data centers for next-generation AI services, Lenovo is developing AI infrastructure around the AGI CPU, and Arm's collaboration with Tsinghua Unigroup also covers both AGI CPU and CSS.

As the Arm server ecosystem continues to expand, the x86 camp, long dominated by Intel and AMD, is facing greater pressure to gain market share. For Arm, the new challenge has shifted from "getting more chips to adopt Arm" to "getting more chips sold directly by Arm itself."

Three-pronged approach

Arm's best business has always been helping others make their fortune.

For years, Qualcomm, MediaTek, Apple, and later, companies like Amazon and Google that moved into the cloud, have been able to design their own chips based on Arm's architecture and IP. Arm doesn't need to predict who will ultimately dominate the market; as long as more chips adopt its technology, it can share in the growth through licensing fees and royalties.

This core business remains highly profitable today. In the first quarter of fiscal year 2027, Arm achieved revenue of $1.289 billion, a year-over-year increase of 22.4%; of which licensing revenue was $574 million, royalty revenue was $715 million, and the non-GAAP operating margin reached 41.2%.

Behind this model, smartphones have always been Arm's largest source of revenue. In fiscal year 2025, smartphone application processors will still contribute approximately 45% of Arm's royalty revenue, corresponding to a market share of 99%, leaving limited room for further growth.

It is against this backdrop that Arm has been looking for room for growth.

AI is clearly the largest growth market right now. In Arm's long-term plan this year, the overall market size for Edge AI, Physical AI and Cloud AI will expand from the current approximately $535 billion to over $1.5 trillion by fiscal year 2031, with Cloud AI being one of the most important sources of growth.

This has also prompted Arm to move its business, which was previously confined to the upstream of the industry chain, further downstream.

From IP to CSS, which has already helped customers complete a large number of system integrations, and then to the AGI CPU, which was launched for the first time in March this year and is ready for mass production, Arm is delivering increasingly complete products to its customers.

In an interview with media outlets such as Wall Street Insights and All-Weather Technology, Arm's Chief Marketing Officer, Ami Badani, stated that this approach largely stems from customer demand. Following the launch of Arm's Computing Subsystem (CSS), an increasing number of customers want Arm to take another step forward and provide products that are even closer to finished chips.

"This is also a major reason why we launched the Arm AGI CPU," Badani said, adding that it represents a further extension of Arm's business model.

At the Arm Everywhere China conference on September 8, Arm laid out almost the entire process of expanding its business boundaries.

In the mobile market, where Arm is most familiar, the company launched the second-generation mobile computing subsystem CSS for Mobile 2, which integrates the new C2 CPU cluster, Mali G2-Ultra NX GPU, system IP, physical implementation solution and software support into a single platform.

Among them, the Mali G2-Ultra NX is the first to integrate a dedicated neural network accelerator in a Mali GPU. Arm is also beginning to extend its cooperation to the application end: vivo has introduced related technologies into its flagship phones, NetEase's "Yan Yun Sixteen Sounds" is promoting the implementation of neural super sampling, and Tencent Games and Unity China are participating in the adaptation from the game platform and engine layer, respectively.

The changes in the cloud are even more pronounced.

The newly released Neoverse CSS N4 is Arm's most configurable generation of cloud-based CSS to date, with a single die capable of supporting up to 128 cores and LPDDR6 and PCIe Gen 7, which is expected to further reduce chip development time.

AGI CPUs bring Arm closer to end-products. This data center-oriented CPU uses a 3nm process, is currently a 136-core design, and supports PCIe Gen 6 and CXL 3.0.

Thus, Arm has established three product roadmaps with varying depths:

Customers with complete chip design capabilities who want to retain more room for differentiation can continue to purchase IP; customers who want to design their own chips and reduce low-level integration work and shorten development cycles can use CSS; customers who value rapid deployment and do not need to design their own CPUs can directly use AGI CPUs.

The three product forms correspond to the different trade-offs customers make between independent design, development cycle and deployment speed, and also further broaden the customer base that Arm can cover.

However, as Arm extends its business to finished chips, the previously simple upstream and downstream relationship has begun to show potential competition.

AWS already has Graviton, Google has Axion, Microsoft has Cobalt, and Nvidia has launched Vera, which is based on the Arm architecture.

In the past, these companies' self-developed CPUs based on the ARM architecture meant that Arm could continue to obtain IP licensing and royalty revenue. However, with the entry of Arm's AGI CPU into the market, competition between the two companies has emerged.

In response, Badani stated during an exchange with media outlets such as All-Weather Technology that AGI CPU is defined as an extension of the market space, and emphasized that IP, CSS, and AGI CPU will continue to coexist for a long time to come.

Badani used SAP as an example, noting that while some of the company's cloud services are currently running on AWS Graviton servers based on the Arm architecture, a large number of on-premises deployments still run on other architecture chips. For Arm, previously accessing AWS via IP did not automatically translate into access to all of SAP's computing environments.

AGI CPUs provide a direct path to these enterprise on-premises deployment scenarios. Arm hopes to use this to extend the Arm architecture to markets that were previously difficult for IP and CSS to reach, while also encouraging more software to adapt to Arm.

The mainland camp is gradually taking shape.

With the rise of the agent boom, the CPU market is expected to expand.

Arm's management believes that as Agentic AI workloads increase, the ratio of CPU to GPU configurations may gradually approach 1:1 in the future.

This is opening up incremental growth potential for CPUs. JPMorgan Chase predicts that shipments of CPUs specifically designed for Agentic AI will increase from 1.6 million units in 2025 to 27 million units in 2028, representing a compound annual growth rate of 155%.

Currently, some mainland companies have already made it clear that they plan to introduce Arm's self-developed chips.

Arm disclosed that Volcano Engine is building next-generation cloud services based on the Arm platform and plans to launch an intelligent agent sandbox solution based on its AGI CPU.

When All-Weather Technology inquired further about the progress of the cooperation at the event, Eddie Ramirez, Vice President of AI Business Expansion at Arm Cloud, stated that the two companies are currently working closely to introduce AGI CPUs into Volcano Engine's data centers and run next-generation AI services on them. A specific deployment timeline will be disclosed by Volcano Engine later.

Lenovo is also developing next-generation AI infrastructure around the AGI CPU. The collaboration between Arm and Tsinghua Unigroup covers both the AGI CPU and CSS, and extends further to agent orchestration, CXL memory pooling, rack-mount system architecture, and edge AI.

However, as the server CPU market expands, the existing competitive landscape is also being rewritten.

The x86 camp, which has long held a dominant position, was the first to be affected.

In the past, the server CPU market was mainly dominated by Intel and AMD. Today, Arm-based server CPUs are entering data centers from more and more directions.

This means that x86 is facing the expansion of the entire Arm server ecosystem.

JPMorgan Chase predicts that Arm's share of the server CPU market may increase from approximately 22% in 2025 to 43% in 2028. Even if the market expansion driven by Agentic AI allows x86 to continue to grow, the market share pressure faced by Intel and AMD may continue to rise.

Competition within the x86 camp is equally fierce. AMD continues to take market share from Intel. JPMorgan estimates that AMD's server CPU market share reached approximately 44% in the first quarter of 2026, and it continues to push forward with next-generation products such as Venice and Verano to compete for new demand in AI servers.

For x86, the immediate challenge is how to maintain market share; for Arm, the bigger challenge is how to continue converting the architectural advantages into chip revenue.

Agentic AI is putting server CPUs back in the spotlight. Arm already has an increasingly large architectural foundation, and the next step is to prove whether it can build a sufficiently large chip business on top of that foundation.

Risk Warning and DisclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.