Arm Holdings has confirmed that it is exploring the development of its own chips, marking a major strategic pivot as it seeks to capitalize on the booming demand for semiconductors powering artificial intelligence and data centers.
The announcement came alongside a weaker-than-expected earnings report that sent the company’s shares down more than 8% in after-hours trading.
Chief executive Rene Haas said Arm is accelerating research and development spending and evaluating the possibility of producing full end solutions, industry shorthand for designing complete chips or chiplets, rather than simply licensing intellectual property.
The move is a major shift for the Cambridge-based company, whose business model has long been centered on licensing processor architecture to some of the world’s largest chipmakers, including Apple, Samsung, and Nvidia.
But amid a changing industry landscape and heightened competition in the AI data center market, Arm, majority owned by Japan’s SoftBank, is seeking to take greater control over its technology and deepen its participation in one of the most lucrative segments of the hardware industry.
“Many of the chiplets that are being developed are mostly Arm IP … and with that we are looking now at the viability of moving beyond the current platform,” Haas reportedly said.
Arm’s ambitions come as its financial performance disappointed expectations. For the quarter ended June, the firm reported revenue of $1.05 billion, up 12% year-on-year but slightly below analysts’ forecast of $1.06 billion.
Royalty revenue, which reflects per-unit sales of chips using Arm designs, rose 25% to $585 million. However, licensing revenue declined 1% to $468 million.
The company guided for revenue in the range of $1.01 billion to $1.11 billion for the current quarter, with earnings per share forecast between $0.29 and $0.37, below the midrange of analysts’ estimates of $0.35.
Arm’s growing interest in chip design could reshape its relationship with long-standing clients, many of whom have relied on its architecture to build their own custom silicon.
Nvidia, a major Arm licensee, uses its cores in AI-focused data center processors. Meanwhile, Amazon and Microsoft have designed custom chips using Cambridge-based firm’s IP to power their cloud infrastructure.
But by moving into chip development itself, Arm risks competing directly with these firms, particularly in the highly strategic area of AI and cloud computing. Earlier reports claim that the firm had already secured Meta as an early customer for its internal chip efforts, underlining the seriousness of its intentions.
SoftBank has made it clear it views the company as central to its AI growth strategy. The Japanese conglomerate is also backing the $500 billion Stargate data center initiative, a joint venture with OpenAI, Oracle, and UAE-based fund MGX, intended to build next-generation compute infrastructure across the US.
While Arm’s shift into end-product development may expand its revenue streams, it also introduces new risks. Unlike its current licensing model, which generates steady income with low overhead, chip design and manufacturing are intensive and fraught with execution challenges.
Rene Haas will also have to tread carefully to avoid alienating major customers who may now see the company as a competitor.
KEY Difference Wire: the secret tool crypto projects use to get guaranteed media coverage