Arm Begins Producing Its Own Chips

Arm, one of the foremost chip design companies globally, announced on Tuesday its initiative to produce semiconductors in-house. This marks a significant shift from its established business model of licensing intellectual property to firms that manufacture and market chips independently. During a live presentation in San Francisco, Arm CEO Rene Haas outlined how the new Arm CPU could enhance the tech landscape and why this is an opportune moment for the company to venture beyond its traditional model and compete directly with other chip manufacturers.

“Let me clarify: We are entering a new phase for ARM, as we will now be supplying CPUs,” Haas stated, showcasing one of the company’s new chips. According to Haas, the primary impetus for this shift is customer demand. Furthermore, as artificial intelligence becomes increasingly prevalent across various sectors, Arm aims to capture a portion of the expanding AI CPU market.

Although speculation around Arm’s in-house chip efforts has persisted for quite some time, the company is finally providing a clearer view of its direction. The new chip is named the Arm AGI CPU, referencing artificial general intelligence—a concept often discussed yet still theoretical—that could rival human capability in various fields. It is engineered to work in conjunction with other chips in high-performance servers within data centers and is intended for agentic AI applications. The chip is being manufactured by Taiwan Semiconductor Manufacturing Corporation, the leading semiconductor foundry globally, and is utilizing TSMC’s 3nm fabrication process.

At the event unveiling the chip, Arm executives highlighted the firm’s legacy of developing energy-efficient chips, asserting that their new AGI CPU will be the “most efficient agentic CPU available.” Compared to rivals such as Intel’s and AMD’s latest x86 chips, Arm claims this new chip will offer superior performance per watt, which measures the energy consumption of computing devices and could potentially save customers billions in energy costs.

The first significant customer for Arm’s new CPU is Meta, which has already received samples. Other companies, including OpenAI, SAP, Cerebras, Cloudflare, and Korean tech firms SK Telecom and Rebellions, have also shown interest in purchasing the chip. Arm anticipates that its AGI CPU will achieve “full production availability” in the latter half of this year.

Santosh Janardhan, Meta’s infrastructure head, joined Haas on stage, expressing his belief that the Arm chip will “expand the [chip] industry on multiple fronts.” As Meta advances towards “personal superintelligence”—AI that delivers deeply personalized experiences—Janardhan mentioned the company’s need for increased silicon, particularly valuing power efficiency.

Kevin Weil, OpenAI’s vice president of science and former chief product officer, also made an appearance alongside Haas. “One of the most frequent requests I hear at OpenAI is: ‘I need more compute,’” Weil remarked. “It’s essentially the currency of our field.”

Nvidia CEO Jensen Huang, Amazon’s senior vice president and distinguished engineer James Hamilton, and Google AI infrastructure chief Amin Vahdat provided pre-recorded video testimonials praising Arm’s latest hardware. Although none explicitly committed to a purchase, all three technology giants utilize Arm’s designs in their processors.

Arm traces its origins back to the late 1970s as Acorn, where it produced microprocessors. In the 1990s, it rebranded to ARM (Advanced RISC Machines), and its then-CEO began licensing chip designs to other businesses. Although the company has since adopted a new branding, his business flourished with the mobile revolution. By the 2010s, many of the world’s major tech companies, such as Apple, Nvidia, Microsoft, Amazon, Samsung, and Tesla, were all leveraging Arm’s technology.