While IoT (Internet of Things) technology has been developed for years, development of AI (artificial intelligence) technology, which took off in 2017, has been quickly integrated with the former into AIoT. At AIoT Ecosystem Forum hosted by Digitimes on June 6 at Computex 2018, speakers all agreed that AIoT will be fast applied in various industries.
Along with widespread application of AIoT, smart sensors and communication networks will be ubiquitous to bring convenience for daily life, said Rahul Patel, Qualcomm Technologies' corporate senior vice president and general manager for Connectivity & Networking Business Unit. However, AIoT-enabled devices should be improved in performance, power consumption and dimension in the future and Qualcomm is making efforts in the direction, Patel noted.
In terms of architecture, IoT systems consist of cloud computing platforms and terminal platforms, with each necessarily being intelligent to some extent to secure smooth operation. For cloud computing platforms in particular, as infrastructure is costly and brand image is necessary for reliability, Amazon, Microsoft and Google have dominated the global market currently, with Amazon Web Services (AWS) being the largest cloud computing platform at a global market share of 33%.
Complete cloud computing platforms are key to smooth operation of AIoT, according to Amazon. Along with mature technological development and increasing successful cases of application, AIoT will be applied in increasingly many industries over the next few years and therefore, enterprises have to select cloud computing platforms that can support multi-industry applications. For example, users of AWS are widely in industries of manufacturing, health care, construction, transportation services, energy and retail.
For AIoT, Taiwan-based enterprises are unable to compete with international IT vendors in cloud computing platform and thus mostly step into AIoT-enabled terminal devices, especially models supporting edge computing.
Patrick Kuo, chairman of ThroughTek - a Taiwan-based developer of smart application of terminal devices - noted that centralized computing entails transfer of data collected at terminal devices to the center, but large volumes of data put heavy burdens on centralized computing platforms and thereby cause relatively slower computing. Edge computing is the best solution of the problem, Kuo said, adding that ThroughTek's long development of image analysis software has proved the advantage for edge computing.
Evolution of AIoT is not only in technology and application concerning cloud computing platforms and terminal ones but also in communication between the two; especially 5G, which is expected to come into commercial use starting in 2019, will bring brand-new business models for AIoT. While 5G is expected to lead to innovative application of IoT systems, telecom carriers should keep trying different business models in a bid to make operation reasonably profitable, Digitimes Research analyst Benson Wu indicated.
As global PC demand is declining, AIoT becomes a new hope for Taiwan's IT industries. With advantage in manufacturing capability, however, Taiwan-based enterprises have to enhance capability in innovation and integrate hardware with AI software.
AIoT Ecosystem Forum hosted by Digitimes at Computex 2018
Photo: Digitimes staff