Arm is expanding beyond chip licensing as it pushes deeper into AI infrastructure and next-generation data centers.
ARM stock has a new mode of operation in this artificial intelligence game, and investors are watching to see if the company can move from its niche licensing business into a more meaningful role as an AI infrastructure company. The company recently reported record quarterly revenue and beat earnings estimates, but the market reacted more nuancedly. Investors are now asking ARM why it isn’t capitalizing on AI growth. Rather, they’re asking if the company can do more than meet that rising demand for agentic AI computing and do it in a way that doesn’t come at the expense of the company’s existing business model, which is primarily based on royalties.
The company’s latest financial results were impressive. ARM generated a record $1.49 billion in revenue for the fourth quarter of fiscal 2026, of which $819 million was licensing revenue, and $671 million was royalty revenue. Non-GAAP earnings per share stood at $0.60, which makes ARM one of the most profitable businesses in the semiconductor industry.
But ARM’s stock got dragged down on the quarterly earnings even though the strong performance was reported, because of the reports that the company has yet to fill all anticipated demand for its new AI chip products. This move moved the spotlight away from the positive earnings surprise, and onto a more pressing thought: how quickly can ARM scale its ambitions around AI to keep up with the growing enterprise demand?
Why Agentic AI Changes the CPU Conversation
Most of the conversation about AI for years has been centered on GPUs, especially those being used to train large language models and generative AI systems. But ARM is asserting that the next generation of artificial intelligence — dubbed “agentic AI” — can significantly alter the significance of CPUs in today’s data center.
Agentic AI is a type of system that can execute actions independently, organize tasks, manage memory, process and execute instructions continuously, and engage and respond dynamically with other AI systems. ARM says these workloads can’t “just be blasted through with raw GPU acceleration. They also rely heavily on CPUs to help them coordinate communication between accelerators, memory systems, networking hardware and software layers.
As AI models become agentic and spread across the world, the company believes that future AI data centres could need more than four times the current capacity of CPUs per gigawatt. That forecast is what ARM hopes to be a CPU market opportunity of at least $100 billion by 2030. The change is significant because it moves CPUs from the level of equipment to one for coordinating the AI infrastructure in the center. As AI workloads shift towards autonomous systems and ongoing inference operations, CPUs might play an ever-more vital role in maintaining the efficiency of large-scale AI environments.
The ARM AGI CPU Is Central to the Strategy
The new ARM AGI CPU platform is the main focus of ARM’s AI growth strategy. The initiative is a more significant step towards producing silicon that is specifically optimized for AI infrastructure, as opposed to the company’s previous licensing model.
The ARM AGI CPU is based on the Neoverse CSS V3 architecture, which features up to 136 Neoverse V3 cores. It features a state-of-the-art 3nm manufacturing process from Taiwan Semiconductor Manufacturing Co., Ltd. (TSMC) and DDR5-8800 memory and high-speed data transfer, making it perfect for AI systems. The particulars indicate that ARM isn’t just looking to be a supplier of intellectual property. Instead, it is looking to be a significant force in the AI infrastructure computing arena.
The company also has a stellar ecosystem for its AGI CPU pursuits. Partners include Microsoft Azure, Meta, OpenAI, NVIDIA, Cloudflare and SAP. The agreements are seen as a sign that hyperscalers and enterprise IT vendors are taking a closer look at ARM CPU deployments for next-generation AI workloads as part of ARM’s broader ambitions for its AI infrastructure. It seems as though Meta has also been collaborating with ARM on multi-generation infrastructure roadmaps, which shows a degree of confidence in the company’s longevity in the field of AI computing.
Record Financial Results Strengthen ARM’s Position
That’s further supported by ARM’s financial results. In FY 26, the revenue was $4.92 billion for the entire fiscal year. The revenue from royalties rose 21% from a year earlier to $2.61 billion and licensing revenue rose 25% to $2.31 billion.
More telling perhaps, ARM said that its revenue from royalties in the data center had more than doubled over the same period last year. This is especially significant, since royalty revenue is thought to be more consistent and of higher quality than single-sale hardware.
The company also guided first-quarter fiscal 27 revenue of about $1.26 billion, in line with Wall Street estimates. While there is concern about the availability of AI infrastructure, management has suggested that demand is still robust. Management said there was strong demand for AI infrastructure, but there were concerns around supply. These numbers are yet another sign of ARM’s advantage with AI expansion, even before its AGI CPU plan gets into full swing.
Supply Constraints Become the Main Risk
Despite the strong financials, supply limitations have emerged as the biggest short-term risk for ARM stock. The company said it has found over $2 billion in customer demand for fiscal 2027 and fiscal 2028. To turn expected demand into real revenue, however, requires making sure manufacturing capacity is in place, advanced packaging, memory supply, and deployment infrastructure for servers are available.
The news that ARM had not booked enough chips to support the extra $1 billion in demand for AGI CPUs prompted investor concerns as soon as they made their earnings call. The issue brings up one of the largest challenges in the semiconductor industry today – the demand for AI is outpacing global production capacity.
This is not an ARM-only issue. GPU manufacturers, cloud providers, and networking suppliers all want the latest in advanced chip manufacturing resources. Still, for ARM, the concern is particularly important because the AGI CPU initiative remains relatively new. Investors want evidence that the company can execute at scale before assigning higher valuations based on future AI growth.
The Neutral Infrastructure Debate
Whether ARM can continue to position itself as a neutral infrastructure provider and still expand its silicon footprint is another key question investors must consider. ARM was a business based on licensing CPU architectures and IP to a variety of semiconductor firms in the past. That neutrality led to ARM technology being widely adopted in smartphones, servers, embedded devices, and infrastructure in the cloud.
But as ARM moves to include more of the AI elements into its full systems, some industry watchers question whether the company has the potential to compete with portions of its own ecosystem. This is a fine balancing act. ARM is seeking to monetize the expansion of its AI infrastructure, but must not risk turning off customers using its IP licensing business. It will be crucial to work closely with hyperscalers, semiconductor designers, and cloud providers as the AGI CPU project continues to grow.
Why Investors Are Watching the Next Phase Carefully
The next stage for ARM stock will depend less on headlines and more on execution. Investors will closely monitor whether the company can secure enough manufacturing capacity to meet rising demand. They will also watch whether data-center royalty growth continues accelerating alongside AGI CPU deployments.
At the same time, ARM must demonstrate that its deeper involvement in silicon products strengthens, rather than weakens, its broader ecosystem relationships. The company’s opportunity is significant. If agentic AI truly drives a major increase in CPU demand inside AI data centers, ARM could become one of the most important infrastructure providers in the entire AI economy.
But the path forward remains challenging. Supply constraints, competitive pressures, manufacturing complexity, and ecosystem management all represent major execution risks.
Conclusion
ARM’s latest earnings report confirmed that the company is already benefiting from the global AI boom. Record revenue, accelerating royalties, and expanding data-center exposure all point toward strong momentum.
Yet the next test for ARM stock is far bigger than a single quarter’s earnings beat. The company now needs to prove that it can successfully scale its AI infrastructure ambitions through the AGI CPU platform while preserving the royalty-driven model that made ARM valuable in the first place.
If ARM succeeds, it may evolve from being viewed primarily as a smartphone architecture company into one of the foundational computing platforms behind the next generation of agentic AI systems.
About The Author
