With related 5G and AI technologies promising significant impacts to various industries, telecom operators not only have to enhance their basic infrastructure but also need to positively undergo transformation enabling them to innovate more value-added services, according to James Lee, vice president of carrier business at Far EasTone Telecommunication (FET).Taiwan's government is now in the process of releasing spectrums for 5G services, and competition for bandwidth is bound to be fierce because only with sufficient bandwidth can telecom operators expand their role in the new mobile communication era and stand firm under the impacts of "5GxAI," Lee said at a recent technology forum held in Taipei recently.The competition is be compounded by the fact that the government is releasing only 270MHz in bandwidth in the core 3.5GHz band through open auction by five local telecom operators, and each operator may need at least 80-100MHz in bandwidth to enable the offering of comprehensive value-added services, Lee noted.Even so, Lee said that he remains positive about the commercialization of 5G networks as Taiwan will be able to leverage its well-established IT ecosystem spanning from IC design, foundry, backend packaging, coupled with profound ICT applications and the availability of related supply chains.Judging from the development of related 5G application services in Korea, Lee continued that remote-monitoring on machinery equipment, collaborative efforts with robots, smart factories, medical care and surveillance will be better choices for 5G services in the early stage.FET VP James LeePhoto: Digitimes staff, December 2019

Three major issues need to be taken into account for the assessment of market trends coming along with 5G: the deployment of mmWave 5G networks, the development of differentiated 5G services, and the feasibility of open networks, according to Benson Wu, an analyst at Digitimes Research.Although the enabling of multi-Gagabit transmission bandwidth is one of the highlights of the Release 15 standalone (SA) 5G specifications approved in mid-2018, the challenge is high for the deployment of mmWave networks for practical applications, leaving the US and Italy the only two countries where telecom operators are able to offer limited regional mmWave service currently, Wu said a recent seminar, "5GxAI Trends," held recently in Taipei.However, there are opportunities for the deployments of mmWave 5G networks in large scale in 2020 as many companies have continued to improve and enhanced related technologies and more countries will release spectrums for use by mmWave networks, Wu noted.For example, Qualcomm has been working on optimizing the design of its mmWave antenna modules, aiming at accelerating the commercialization of mmWave networks.The planned release of mmWave 5G frequency bands by a number of nations including Russia, Australia, Singapore, Taiwan and others in the EU in 2020 will also help push the deployments of mmWave networks.While eMBB (Enhanced Mobile Broadband), mMTC (massive Machine Type Communications) and URLLC (Ultra Reliable Low Latency Communications) are three sets of use cases defined for 5G, commercialization of mMTC and URLLC services will lag behind that of eMBB's, Wu opined.However, judging from the fact that the average revenue per megabyte (revenue per MB) has continued declining in recent years, the development of 5G eMBB applications has to be further differentiated so as to improve the value of data traffic, Wu commented.Based on the standards set for slicing networks, 5G networks will gradually move towards virtualization, and this trend will bring new opportunities for white-box equipment providers to enter the traditional telecommunications equipment market, Wu asserted.Digitimes Research analyst Benson WuPhoto: Digitimes staff, December 2019

High-speed transmission, wide coverage and low latency of 5G technology are expected to be integrated with AI cloud computing platforms and edge computing architecture for application to smart city, transportation, medical care and manufacturing in 2020, a consensus from experts attending a seminar jointly hosted by Monte Jade Science & Technology Association of Taiwan (MJSTA Taiwan) and Digitimes.South Korea is the first country to start 5G commercial operation around the world, with the number of 5G users reaching nearly 3.5 million as of the end of September 2019, exceeding the originally estimated two million for the end of the year - a sign that global commercial use of 5G will take off in 2020.5G and AI are not two unrelated technology sectors, as they capitalize on each other's technological features to create synergy, MJSTA Taiwan chairwoman Lee Jih-chu said in her speech at the seminar.Based on Taiwan's technological strength in ICT and semiconductor, Taiwan-based enterprises stand a good chance of sharing "5GxAI" business opportunities, Lee noted. However, Taiwanese enterprises have to cope with unfavorable factors concerning trade protectionism arising from anti-globalization in some developed countries, and MJSTA will help them enhance their international competitiveness, Lee indicated.MJSTA Taiwan is a comprehensive science & technology platform headquartered in Taiwan and has 16 overseas branches, Lee said.Drastic changes in the industrial structure caused by AI and cloud computing, and global impacts from the US-China trade disputes give Taiwan-based enterprises good opportunities, but the problem is whether they can capitalize on the opportunities, said Digitimes president Colley Hwang at the seminar.For 5GxAI in particular, Taiwan-based enterprises have technological advantages supporting their international competitiveness, Hwang noted.As each area of 5G or AI application, such as medical care and transportation, involves intensive domain knowledge, applications should be based on DFSI (domain focus system integration) concept, Hwang indicated.JMSTA Taiwan chairwoman Lee Jih-chuDigitimes president Colley HwangPhotos: Digitimes staff, December 2019

While 5G and AI will dominate new emerging tech applications to create cash flows for enterprises in the next decade, relevant semiconductor materials and technologies will not only serve as a market enabler but also affect the positions of players in the semiconductor market, according to Tony Huang, deputy director at Digitimes Research.Huang, speaking at a "5GxAI Trends" seminar held recently in Taipei, said 5G base stations and handsets will rely on higher-end RF front-end modules and power amplifiers to meet requirements for increasingly sophisticated and efficient communication performances, with more-advanced semiconductor materials badly needed to suppor them. Significant upgrades on AI-based edge computing capability are expected to materialize in the next few years to help broaden the scope of AI applications. All this has indicated that semiconductor technologies will become even more important.Huang said that Moore's Law remains applicable to the advancements seen in handset APs for iPhones and networking chips by Broadcom, while the latest product development roadmaps of leading foundry TSMC and major semiconductor equipment supplier ASML also suggest that IC scaling technology will continue to advance in the next decade, with 2nm node to be available in the near future.To reduce IC fabrication cost, heterogeneous integration of diverse chipsets via 3D IC, SiP packaging technologies and more-advanced processes will continue to gain ground, according to Huang.He also noted that the volumes of transceiver filters, RF switches and PA devices needed for 5G handsets will grow exponentially, compared to those for 4G models, and mmWave 5G smartphones also will need more RF devices than sub-6GHz models, stoking demand for AiP (antenna in package) technology that can avert mmWave signal interference.Third-generation compound semiconductors will be also increasingly applied to 5G base stations, as III-V semiconductor materials, especially GaN, can facilitate construction of transmission systems smaller in size and lower in power consumption and cost, Huang indicated.The seminar was jointly organized by Monte Jade Science & Technology Association of Taiwan and Digitimes.At the event, many tech firms also showcased their latest 5G and AI solutions, including Han Tai Technology presenting graphene materials featuring high heat conductivity, Chang Chun Group displaying high-frequency, high-speed copper foil ideal for 5G high-speed communication needs, Xander International exhibiting multi-tasking wireless communication chips modules, Innodisk demonstrating high-end memory solutions for industry control use, Vicor offering 48V power management chips, and Aver Information highlighting smart video conferencing systems, among others.Digitimes Research deputy director Tong HuangPhoto: Digitimes staff, December 2019

Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP) have offered AIaaS (AI as a service) products to promote application of AI technologies through reducing technological barriers for developers of AI solutions, according to Digitimes Research analyst Joyce Chen.Chen was speaking at a "5GxAI Trends" seminar jointly hosted by Monte Jade Science & Technology Association of Taiwan and Digitimes.Voice assistants are a typical application of AI to daily life, Chen said. Smart speakers, for example, involve upstream technologies in the AI supply chain, including basic computing, data/information and machine learning frameworks, and mid-stream technologies such as common voice technology, with cloud computing platforms playing a key role in the AI supply chain, Chen noted.AIaaS products offered by AWS, Azure and GCP mainly consist of edge AI tools used in IoT, AutoML (automated machine learning) and MarTech (marketing technology), Chen said.Edge AI tools, such as AI cameras, help users to test AI solutions developed themselves to shorten time to market, Chen indicated.AutoML automates the process of applying machine learning to solving problems through replacing labor-consuming repeated work entailed in machine learning modeling, Chen noted. For example, Amazon SageMaker simplifies the process of machine learning modeling; GCP GUI (graphical user interface)-based AutoML services exempt users from any encoding; and Azure AutoML services enable users to see effects in training of machine learning models, Chen said.MarTech, such as precision marketing, uses AI to generate the most commercial value from using operating data to hike business performance, Chen indicated.Digitimes Research analyst Joyce ChenPhoto: Digitimes staff, December 2019