Arm Holdings on Tuesday announced plans to design and sell its first in-house silicon product, marking a significant shift from its decades-long business model centred on licensing chip designs.
The new processor, called the AGI CPU, is aimed at powering artificial intelligence workloads in data centres and is expected to open up a multi-billion-dollar revenue opportunity for the company.
The move represents a major strategic pivot for Arm, which since its founding in 1990 has largely avoided manufacturing chips, instead supplying intellectual property to semiconductor firms.
A ‘pivotal’ shift beyond licensing model
Arm has historically generated revenue by licensing its chip architectures to companies such as Qualcomm and Nvidia, collecting royalties based on unit shipments.
This asset-light model helped it become one of the most influential players in the global semiconductor ecosystem, with its designs powering the vast majority of mobile devices.
The introduction of its own chip signals a departure from that approach, bringing Arm into more direct competition with some of its own customers.
Chief executive Rene Haas described the development as a pivotal moment for the company, as it looks to capture a larger share of value in the rapidly expanding AI infrastructure market.
Analysts said the move could reshape Arm’s long-term growth trajectory.
Pierre Ferragu of New Street Research called the shift to selling chips the most significant strategic pivot in the company’s history.
Targeting next-generation AI workloads
The AGI CPU is designed for a new class of artificial intelligence applications known as AI agents, which can perform tasks autonomously with minimal human intervention, rather than simply responding to prompts like traditional chatbots.
These workloads require substantial computing power and energy efficiency, areas where Arm is seeking to differentiate its offering.
The chip features up to 136 cores and is designed to deliver a significant increase in performance per watt, a key metric for large-scale data centre operations.
Mohamed Awad, head of Arm’s cloud AI business, said the processor could dramatically reduce infrastructure costs.
He estimated that adopting the technology could save up to $10 billion in the construction of a large AI data centre, which can cost as much as $50 billion.
Strong industry backing and partnerships
The development of the AGI CPU has drawn support from major technology players.
Meta, the parent of Facebook, collaborated on the chip and has signed on as its first customer.
OpenAI is also among the early adopters.
Manufacturing will be handled by Taiwan Semiconductor Manufacturing Company using its advanced 3-nanometre process technology.
The chip is built using two separate pieces of silicon that function as a single unit.
Arm said it has already received working test chips and plans to begin volume production in the second half of the year.
In addition to the chip itself, the company is collaborating with hardware partners such as Lenovo and Quanta Computer to deliver complete server systems.
Competitive landscape intensifies
Arm’s entry into chip production comes at a time when demand for AI hardware is surging.
Nvidia has dominated the market for specialised AI processors, benefiting from the rapid adoption of generative AI technologies.
However, attention is increasingly shifting toward general-purpose processors that support broader AI workloads, particularly as companies develop more advanced AI agents.
Nvidia chief executive Jensen Huang has previously highlighted the growing importance of such chips, suggesting they could become a major revenue driver.
Arm’s new offering positions it to capture a share of this emerging market, while also deepening its role in the AI ecosystem.
What analysts feel about the pivot
Wall Street currently expects Arm to report revenue of about $4.91 billion and earnings of $1.75 per share for the current fiscal year, according to estimates compiled by LSEG.
Some analysts are increasingly optimistic about the company’s prospects.
In a rare move, HSBC on Friday double upgraded the stock, citing its transition from a smartphone-dependent business to a key player in AI server processors.
It raised its rating on the stock to ‘Buy’ from ‘Reduce’ and more than doubled its price target to $205 form $90, with the $205 mark being the highest on Wall Street.
“We believe Arm is now firmly in the middle of a transition from being a smartphone-dependent semi-IP play, into a major AI server CPU beneficiary that remains undervalued by the market,” wrote HSBC analyst Frank Lee.
Lee said royalties from server CPUs could eventually match Arm’s current total revenue, potentially reaching $4 billion by 2030.
He also noted that if Arm successfully establishes itself in the merchant CPU market, it could significantly increase average selling prices and improve profitability.
However, challenges remain.
The global smartphone market, a traditional stronghold for Arm, has been under pressure, raising concerns about near-term growth.
Bank of America has maintained a more cautious stance, pointing to uncertainties around execution and the potential risks of entering a capital-intensive manufacturing business.
It has reiterated a Neutral rating and a $140 price target in a research note on Thursday.
The post Arm pivots to chipmaking with AI chip launch: what this could mean for ARM appeared first on Invezz
