The world is full of analog signals and remain so. Where as present digital electronics computing world is full of digital signals except at the input and output interface to the outside world. Compared to how we human process data, the digital electronics process it in a weird way with lot of number crunching. With the arrival of artificial intelligence (AI) computing, combination of both analog computing and digital computing to play a role in future systems.
Analog role in future computing to increase and to become key diffrentiator. In that way Analog IC design to become more important not only as front-end signal chain processing but more of analog computing. Thousands of ADCs and DACs, analog comparators and some form of analog data processing and storing tech to drive the technology growth. Just like the brain handle a picture data, each pixel to get analyzed at the stored memory by using a concept called computing in memory. There are startups exclusively working in this area. Picture recognition, voice recognition to leverage this technology in offering real time recognition of images and voice in a smaller semiconductor at lower power and with less latency.
The world is analog and much of the reading and analyzing is done through simple matching of analog signal. Digital involves conversions and computing loops, where as analog is direct. Can analog circuit be fed with two linearly varying inputs and if it can sum them accurately that is lot more simple, faster and better than a digital full adder based summing circuit. Though it looks simple in concept, designing such analog circuit is least explored to perform computing.
Multiply-accumulate ( MAC) or Multiply-add ( MAD) operation is a digital process used in digital signal processing. Now it is more and more used in AI processors for inference purpose. These are called deep neural networks. Analog circuits make up the MAC and MAD and any such stream data processing circuit. Flash transistors are used to model variable resistors. Analog to digital converters convert current to digital codes. A startup company called Mythic is extensively into this type of computing. Mythic calls it mixed signal computing.
Irrespective of computing opportunities for analog integrated circuits, there are lot of normal analog usage application market opportunities in the growing automotive and industrial automation sector, where increasing number of sensors are connected to digital systems.
The analog IC manufacturing is highly processed driven, need lot of fine tuning at each step of the process to achieve yield. Lot of Ips are involved. It's not easy for a start-up to get into analog IC fabrication market. Fabless model does not work that well for analog IC fabrication.
The recent market research by IC Insights on this key semiconductor market reveals: The analog semiconductor market grew by 30% in 2021, while the total IC market increased 26% in 2021. IC Insights further says every general purpose and application specific analog product segment seen double-digit sales growth in 2021. Signal conversion revenue grew 13% last year, but revenue in every other analog IC category jumped by at least 27%, as per IC Insights.