Artificial intelligence (AI) has been a dream of humanity for centuries. Yet it has not been until recent progress in machine learning that it has been possible for AI to solve real-life problems. GlobalData predicts that this rapid technological progress and capability will drive global artificial intelligence (AI) chips revenue from $12 billion in 2021 to reach $130 billion by 2030, at a compound annual growth rate (CAGR) of 30%.
Josep Bori, Research Director at GlobalData Thematics Intelligence, comments, “This rapid expansion will be driven by chips specifically optimized for AI with their share of the combined micro-component and digital logic semiconductor market set to increase from less than 10% in 2021 to at least 40% by 2030.”
Bori continues, “Deep learning neural networks continue to expand their capabilities, now including face recognition, medical diagnosis, and self-driving cars. This has been led by an improvement in the mathematical models used and the exponential growth in the model sizes and training data sets.”
The popular OpenAI’s GPT-3 model, which is already capable of writing original prose with fluency equivalent to that of a human, has 175 billion parameters and takes months to train. GlobalData expects the number of real-life use cases to continue to grow, particularly in the self-driving cars and robotics arenas, which will further show the limits of traditional processing semiconductors and drive demand for AI chips.
Currently, most of the progress in the development of AI chips is coming from redesigning classic microprocessor architectures to optimize instruction sets for linear algebra, use compact data types, and enable massive parallelism and memory on the chip. However, we should not forget the disruptive potential of radically innovative computing architectures such as neuromorphic chips as well as quantum computing. The microcomponent and digital logic market are growing 7% annually however the AI component of this market is achieving over 40% growth.
Bori adds, “In our view, the ongoing trade dispute between China and the US has negative implications for the global progress of AI semis technology. We believe China will play a leading role in AI, due to its leadership in AI software and IoT technology, and its progress in low-end chip manufacturing. However, unless China solves its access to extreme ultraviolet (EUV) lithography technology, currently indirectly prevented by US sanctions, it will likely struggle in AI in the datacenter, and most likely autonomous vehicles.”