ARM, long called the king of semiconductor blueprints, has begun developing chips itself. The industry sees this not as a simple business expansion but as a shift from a "design IP supplier" to a system platform company for the artificial intelligence (AI) era. Analysts say the move reflects a judgment that, in the AI era, relying on IP (blueprints) alone makes it difficult to secure both profitability and leadership of the ecosystem.
Until now, ARM maintained a model of not producing chips directly and licensing central processing unit (CPU) designs (IP). Global corporations such as Apple, Qualcomm, Nvidia, and Samsung Electronics developed their own chips based on ARM designs.
However, as the AI market surges, this structure is showing its limits. AI computation has made optimization of the entire system—combining the graphics processing unit (GPU), Neural Processing Unit (NPU), high-bandwidth memory (HBM), and more—rather than single CPU performance, the core competitive edge. In this process, companies that only supply designs have remained with limited revenue, while those that control both chips and software have come to capture most of the added value.
Nvidia is at the center of this trend. Nvidia is leading the AI ecosystem through a platform strategy that combines the GPU and CUDA. Reports indicate that inside ARM, there was a conclusion that it would be difficult to keep up with this competitive landscape through the design business alone. Accordingly, ARM is expanding its strategy beyond reference designs to computing subsystems (CSS) and in-house chip development.
These changes are also affecting relationships with existing clients. In the past, there was a cooperative structure between design providers and manufacturers, but as ARM enters the chip market directly, a competitive dynamic is emerging. Analysts say the burden has grown especially for Qualcomm and Samsung Electronics, which are highly dependent on ARM. If ARM develops its own chips, clients could see overlaps in design direction or face disadvantages in technical support. In response, Qualcomm is accelerating development of its own CPU core, "Oryon," and Samsung Electronics is also reviewing ways to reduce its dependence on ARM by strengthening its independent design capabilities.
ARM's strategic shift also appears driven by the task of securing growth after its listing. ARM, listed under SoftBank, is in a position to present investors with new growth engines. While the IP business provides stable revenue but has limited growth, the chip business carries significant risk but is seen as an area that can sharply boost corporate value if successful.
ARM's strategy, which declares itself a "chip maker," is a step toward transitioning from a CPU design company to an AI platform company. The move aligns with Nvidia's platform strategy and Apple's vertical integration. As competition in the AI era is decided not by individual chips but by entire systems, the corporations that seize the platform are taking the lead. An industry official said, "If you remain an IP designer, you could lose both revenue and leadership in the AI era," adding, "ARM's move is an inevitable gambit to break through this, and the beginning of an ecosystem-level evolution faced by semiconductor design asset (IP) corporations."