Unveiling the Future: The Rise of AI Chips in the Tech Industry

At its Ignite developers’ conference, Microsoft made a groundbreaking announcement by unveiling chips specifically designed for AI computing tasks. This move signifies a shift away from traditional CPUs towards specialised processing units tailored for executing AI models. In a parallel development, Qualcomm and MediaTek have also joined the race by introducing on-device generative AI capabilities in their upcoming chipsets for smartphones.

The introduction of AI chips marks a significant advancement in the semiconductor industry, with a focus on enhancing on-device AI capabilities, particularly in executing Large Language Models (LLMs). These chips, typically configured as ‘system-on-chip’ (SoC), go beyond the capabilities of general-purpose CPUs by incorporating functions specifically optimized for AI tasks.

To understand the need for dedicated AI chips, it is essential to grasp the practical workings of AI. When your smartphone camera identifies a dog, it relies on AI algorithms trained to recognize specific objects. This training occurs within a neural network, either on the device or in the cloud, mimicking the human brain’s decision-making process.

Unlike traditional CPUs, AI chips are designed to handle parallel computing, enabling them to execute multiple calculations simultaneously. This parallel processing approach results in faster and more efficient performance, making AI tasks more manageable on devices. GPUs, while capable of handling such workloads, are not specifically designed for AI tasks, necessitating the development of dedicated AI chips.

Various types of AI chips serve different purposes, with GPUs being used for algorithm development and refinement, FPGAs for real-world data inputs, and application-specific integrated circuits offering flexibility for both training and inference tasks. The unique design features of AI chips contribute to their superior speed and efficiency compared to CPUs, ultimately enhancing the training and inference of AI algorithms.

The emergence of AI chips from Microsoft, Qualcomm, and MediaTek signals a significant industry trend towards specialized processing units for AI tasks, paving the way for more advanced on-device AI capabilities in smartphones and other devices. This shift underscores the importance of dedicated hardware in enabling efficient and effective AI computing.