Arm is set to launch its first self-designed chip this year, marking a major departure from its usual business model of licensing chip blueprints to companies like Apple and Nvidia. The company’s CEO, Rene Haas, is expected to unveil the new chip by summer 2025. This shift to creating complete processors could disrupt the $700bn semiconductor industry and put Arm in direct competition with some of its largest clients.
Meta has already signed on as one of the first customers, and Arm’s chip will be designed primarily for use in large data centers, with customizations for clients like Meta. The production will be outsourced to companies like Taiwan Semiconductor Manufacturing Co.
This move is part of a larger strategy by SoftBank, Arm’s parent company, to expand its presence in AI chip production. SoftBank’s founder, Masayoshi Son, has been pushing Arm toward creating more value from its intellectual property, with the long-term goal of contributing to AI infrastructure, which includes the $500bn Stargate initiative involving OpenAI, Microsoft, and Nvidia.
Additionally, SoftBank is nearing a deal to acquire Ampere, a chip design company valued at $6.5bn, which is key to supporting Arm’s move into chip manufacturing. As part of its growth, Arm’s value has surged to $160bn since its 2023 Nasdaq listing.
Meta’s involvement with Arm is part of its effort to enhance efficiency for AI tasks by using custom-designed chips, gradually replacing Intel and AMD’s offerings. Arm’s power-efficient designs, already widely used in mobile devices, are becoming an attractive alternative for AI-driven data centers, which require more energy.
The decision to build its chips could also bring Arm into conflict with major clients like Qualcomm and Nvidia, raising concerns about potential legal disputes over licensing terms.
No Comments