AI, in the palm of your hand... ARM-led "Heterogeneous Computing" Revolution

AI technology is increasingly expanding into everyday devices, and the core technology leading this transformation—“Heterogeneous Computing”—is on the rise. This is a concept that enables units with different performance and functions, such as CPU, GPU, NPU, to work together organically within a single system, while achieving performance optimization and energy efficiency. Chris Burke, Senior Vice President of ARM’s Edge AI Business Division, emphasized at CES 2026: “We are currently in a transitional period where AI is moving away from data centers and into user devices,” and pointed out that such technologies are creating perceptible differentiated experiences in daily life.

ARM is leveraging a strong collaboration with NVIDIA to enable AI-driven capabilities on various edge devices such as smartphones and wearables. Burke explained, “Smartphones containing Arm CPUs, GPUs, and NPUs are typical representatives,” and “thanks to heterogeneous computing, users can experience faster, smarter AI devices with longer battery life.” He specifically noted that by utilizing compute units optimized for specific functions, it is possible to break through the limitations of traditional architectures that rely on a single processor for all computations.

This architectural innovation not only targets high-performance AI computations but also fundamentally changes the software design paradigm for developers. Burke mentioned, “Effective programming techniques to utilize GPUs or NPUs are key, and ARM is investing heavily in this area.” He explained that the industry is moving away from the previous approach of adding AI capabilities retroactively, toward engineering practices where AI is integrated as a fundamental element from the initial design stage, and this is becoming increasingly common.

Meanwhile, wearable devices are becoming a new stage for testing the performance of combining heterogeneous computing and AI. Sensor-based personal devices can process personalized data in real-time, and their advantages in privacy protection and response speed are increasingly evident. Burke stated, “Wearables are evolving from simple peripheral devices to platforms where AI runs directly on the body,” and “their influence is expanding into health management, security, and even quality of life improvements.”

However, technological progress alone cannot solve all problems. He emphasized, “Real AI devices must have the same level of reliability and security as high performance,” and pointed out that security features must be built into the architecture layer. Especially as edge AI handles sensitive information, trust-based design is becoming increasingly important.

At the core of this shift in AI dominance—from data center-centric to user-centric—is heterogeneous computing. ARM is seizing this opportunity to change the landscape across the entire ecosystem, including new hardware designs, developer experiences, and consumer products. The foundational technologies that make AI smarter, more accessible, and more secure are now becoming a reality.