The rise of artificial intelligence (AI) has led to the development of specialised chip designs with innovative features to manage computationally intensive tasks and provide critical power management capabilities. Chip designing is an extremely time-consuming and complex process involving various business, environmental, and operational considerations in chip innovation to meet the evolving demands of end-markets by striking the balance between performance, efficiency and cost. Furthermore, today chip design is expected to be carried out at a rapid pace to match the pace of innovation in end-markets and use-cases spurred by AI.
The emergence of advanced AI applications is acting as a catalyst for the development of advanced and sophisticated AI chips. In this blog, we will provide a brief overview of the AI chipsets market (which has been discussed in detail in our ‘AI Chipsets: cloud AI drives scale, AIoT drives innovation’ report), with a focus on cloud chip IP and a brief introduction to a range of AI chipset vendors.
The AI chips ecosystem for chips that are destined to be deployed in cloud data centres is evolving with the rising demand for AI chips to process and handle huge volumes of data in the context of application-specific tasks. Despite this growth, the industry remains dominated by a few key players. While new ventures are emerging slowly, semiconductor giants still control most of the market across the entire value chain.
The chip market is dominated by the US, China, and Taiwan, each of which controls significant aspects of supply chains for semiconductor chips. The US leads in Electronic Design Automation (EDA) and design, while Taiwanese companies excel in the manufacturing of chips. China and South Korea hold a smaller, yet significant share in chip manufacturing. Japan and Taiwan dominate the supply of high-quality silicon wafers for AI chip production. In terms of specialised machinery for design imprinting, the Netherlands-based Advanced Semiconductor Materials Lithography (ASML) is the sole provider of extreme ultraviolet lithography machines, crucial for imprinting circuitry designs onto silicon wafers for the most advanced AI chips. Overall, the chip industry is driven by "American-East Asian" companies, together supplying the most powerful chips worldwide.
Large datasets are frequently processed by using AI processing in cloud datacentres, which often demand high computational power. As traditional CPUs lack crucial characteristics of parallel processing, advanced and sophisticated AI-optimised chips are being developed to process billions of operations per second. Moreover, providers of AI-optimised chips for such datacentres must also ensure minimal energy consumption to be able efficiently to offer high computational power, so that datacentre infrastructure can be equipped with adequate cooling systems for heat management.
Multiple established as well as new companies are venturing into the Cloud AI focused chips market. Discussed below are some of the key vendors.
The AI chipsets market is dominated by companies such as Nvidia and TSMC, together providing chips for data centres to support compute-intensive tasks such as training and running large language models and generative AI. In this fast-growing market, the market lead of the incumbent players in this space seems almost unassailable.
