i卡會員

COVID-19 has changed the way of life for many people during the past two or more years. Whilst countries around the world are gradually emerging from the pandemic, the wide range of convenient and remote applications that have arisen from the pandemic will continue to thrive.   

New software has contributed to a change in users’ behavior. For example, the need for remote working during the pandemic has resulted in a significant increase in the downloading of online conference apps such as Hangouts, Meet, Microsoft Teams and Zoom. Social networking services such as Houseparty also experienced a rapid growth in the number of downloads.

The use of virtual conferencing systems such as Spatial Systems for work and learning has been increasing. Amazon AR Shopping and NextVR are also seeing higher demand driven by the growth of online entertainment and social networking such as AR shopping, VR gaming and game watching. 

Meanwhile, to avoid the overloading of the medical system caused by people rushing to hospitals for any health concerns, the market for wearable devices to monitor the wearer’s physical status at home is on the rise. The adoption of products such as the Kinsa Smart Thermometer and the Fitbit Charge 4 for tracking blood oxygen levels has been increasing. Home fitness devices such as Smart Mirror and connected fitness equipment are gaining popularity possibly because people have been unable to go to the gym. A large number of innovative Internet-of-Things (IoT) products are finding ways to become part of our day-to-day life at an ever-quicker pace. This is a new business opportunity for end-consumer products resulting from the pandemic. 

Other innovative use cases such as autonomous transportation and smart unmanned factories are also emerging. This will expand the adoption of advanced semiconductors in a variety of devices. The combination of 5G and AI requires much faster processing speeds than before, whether in the cloud or for edge computing. 

Al chips previously used in cloud servers are seeing higher costs and electricity consumption due to the growing requirement for computing functions and expanding data volumes. As a result, more and more edge computing devices will be deployed in our daily life. Edge computing devices with AI chips will be prevalent at retailers and distributors and in industries such as healthcare, manufacturing, transportation, smart cities and related infrastructure.

It can be expected that the sensors installed at factories, in cars and in the retail and service industries will collect a large amount of data for processing by edge computing. The use of cross-referencing can reduce data processing and application costs and hence improve service quality.

The requirements for edge computing are rapidly changing. The focus of chip R&D is on the improvement of chip performance and the reduction of energy consumption and transmission latency to enhance system reliability. With AI chips, edge computing will integrate the functionality of cloud servers to serve as the foundation of hybrid computing. This is a key area of focus for Taiwanese manufacturers.

5G and AI are accelerating the growth of the semiconductor industry. However, complex computing makes power consumption a bottleneck. The Department of Industrial Technology of the Ministry of Economic Affairs has been leading the Industrial Technology Research Institute (ITRI) in moving forward with the development of in-memory computing and embedded memory, and in cooperating with domestic semiconductor companies and the University of California, Los Angeles (UCLA). With support from the Defense Advanced Research Projects Agency (DARPA), the program is striving to develop the next generation of memory, which will be faster, more reliable and energy efficient, and able to retain data when the power is turned off. 

It is envisioned that the performance will be comparable to that of Intel and ahead of Samsung, and it will overcome the interference problems of memory arrays, which will be a huge step forward for the industry. If used in mobile phones, the power consumption for AI functions could be extended from one day to over three days - a new milestone. In the future, heterogeneous packaging technology will enable the integration of more advanced processors. The outlook is extremely bright for AI, auto chips and high-performance computing chips. 

Meanwhile, the Department of Industrial Technology has established the AI-on-Chip Taiwan Alliance (AITA) by working with domestic semiconductor companies to set up four technology special interest groups (SIGs). AITA now has a total of 147 domestic and overseas members. AITA members discuss the direction of the technology roadmap on this platform. Leading domestic semiconductor companies also provide chip R&D and formulate common technology interfaces to accelerate the R&D of next-generation semiconductors. It is hoped that the collective efforts by the industry, academia, research organizations and government agencies can create the next economic miracle for the Taiwan semiconductor industry.