Abstract
As chipmakers design different types of chips to enable Artificial Intelligence (AI) applications, the adoption of AI chips has increased recently. To support applications based on deep learning, AI chips have inbuilt AI acceleration and are created with a specialized architecture. One of the key drivers boosting the market's expansion is the increasing integration of AI processors in data centers. The major significance of using AI chips when compared with traditional ICs are fast computational integration and large bandwidth. This study summarizes the need of the AI chips and its functionalities and how the AI chips varies from the general ICs. Finally, a discussion on the potential AI chip initiatives are provided.