TAIPEI, Nov. 8, 2022 – This article is based on an interview undertaken by FusionMedium’s technology online media, TechOrange, and published with permission:
When Google and Apple start making their own chips, what kind of paradigm shift will this mean for the global semiconductor industry? In recent years, the high-tech manufacturing market has been making efforts to respond to the demand for diversified smart devices such as cell phones, home appliances, and smart cars. With an increasing demand for customized AI chips, such components cannot be produced in traditional supply chain models.
Both edge and environmental computing also require AI chips with high efficacy, low power consumption, and high computing power to solve issues related to data privacy, power efficiency, low latency, and computing power. According to ABI Research, a leading technology industry market research firm, the size of the edge computing chip market is expected to reach US$12.2 billion in 2025, surpassing, for the first time, that of the cloud AI chip market.
To meet the needs of tomorrow, Taiwan AI chip start-up Kneron has been developing its own AI system on monolithic chips since 2015. The firm has already established a presence in the United States, mainland China, Taiwan, and Northeast Asia. Albert Liu, Founder & CEO of Kneron, said that since the first day of operation, Kneron has stood by its self-stated mission of “making AI ubiquitous”, and expects that in the future, for every terminal device – whether an Edge Server, an IoT device, a household application, security control or automotive component – Kneron will be the provider of the intelligent AI chips that run it.
How does Kneron take the lead in the AI market? CEO Albert Liu: Fast, early and hard-working
When asked how Kneron has been able to move from cutting-edge technology research and development to product launch in just 7 years to become a leader in the security and automotive industries, and how the company developed its business logic, Liu modestly smiled and said, “There is no trick to it. We are just faster, earlier, and more determined than others.” Compared with the world’s major manufacturers, Kneron realized the importance of implementing applications early. So the chip maker started to focus on developing products and customers from the time when the concept of edge AI was still new and yet relatively unknown. As the technical impact of edge computing applications in the market gradually improved, Kneron, with its advantages as a pioneer, knew how to position its products and better meet the needs of customers in the market.
“In the process of AI industrialization, it is also important to know the characteristics, acceptance, and timeline of each market,” Liu said. Since the Chinese market is highly receptive to new things and is open to the quick introduction of new products, in 2016 Kneron partnered with Chinese Internet companies Tencent Holdings, Baidu and Sogou to roll out a pure cloud-based AI algorithm service, and, by doing so, acquired a large amount of car data and gained a head start in the industry. Kneron went on to secure funding from several venture capital firms in Europe, the US, and Asia. After that, in 2017, the company started to expand into Japan, South Korea, Europe, and the United States.
Now in its seventh year, Kneron’s product lines are divided into four major categories. In the server market, customers include Quanta Computer, Chunghwa Telecom, Comcast Corporation, and Naver, Korea’s largest portal. In the Internet of Things segment (including home appliances and security control), customers include Hanwha group, D-Link, and Italy’s Spark Security. In automotive, customers are mainly Chinese and Japanese brands, including Toyota and Nissan, among others. In recent years, Kneron has also partnered with MIH Consortium and Zhongxing Bus.
Identifying the demand of a decentralized architecture for the vehicles of the future, Kneron enters the automotive market with “Lego-style” chips
Liu said that the reason that Kneron decided to enter the automotive field in 2020 is closely related to the industrial introduction cycle. The introduction cycle of the Internet of Things (IoT) sector is short and fast, which, from R&D to putting the final touches on the product and then on to mass production, is about 8 to 9 months. However, the introduction cycle of the automotive pre-loading sector, from the start of a collaboration to actual mass production, often spans 3 to 5 years. Currently, vendors are few that can accommodate such a long partnership cycle and deliver AI, chip and system solutions.
In order to meet the needs of application-side customers, Kneron’s strategy is to develop vertical product lines – software, applications, models, and underlying chips – in a single package that can be adjusted simultaneously, so as to gain a head start in the AI era and earn the trust of key multinational customers with practical applications, demonstrating its flexibility and flexible differentiation advantage.
What are the future challenges for Taiwan’s automotive supply chain?
Taiwanese manufacturers want to start solving the challenges faced by the automotive market with AI chips. Commenting on this, Liu said frankly that the innate lack of a complete automotive industry structure is Taiwan’s biggest challenge. “If I were an IC factory in Taiwan today and wanted to make automotive chips, the first issue that I must contend with is that Taiwan does not have its own automotive industry and a brand with an enough market share. So, upstream and downstream integration, data collection, iteration and actual implementation would all be quite difficult.”
Liu said this is why Kneron still generates 70-80% of its revenue from overseas markets. In addition to gradually strengthening and expanding the competitiveness of the Taiwanese market, Kneron has been, on the basis of its global portfolio that it established immediately upon starting operations, directly connecting with the world in terms of both talent and market, including recruiting seasoned engineers and executives from Taiwan, mainland China and the United States, and taking advantage of their strengths in semiconductor processes, wide and fast-growing markets, and cutting-edge technologies, in an effort to open up a wider range of possibilities for the company’s growth path.
In many countries, the development of AI chips is highly correlated with government and industrial policies. In addition to expanding into international markets, Kneron is actively cultivating the local AI ecosystem in Taiwan, including promoting AI-related education courses in rural elementary schools, high schools, and universities, as well as joining the AI on Chip Industrial Cooperation Strategic Alliance launched by the Smart Electronics Industry Project Promotion Office (SIPO) of the Industrial Development Bureau of the Ministry of Economic Affairs, where he and his colleagues were able to gain an in-depth understanding of industry trends through the services provided by the Alliance, such as industry-specific business opportunity seminars, and industry promotion and exchange events. The Alliance also helped introduce partners from the upstream and downstream of the industry chain to jointly invest in the development of core AI on Chip areas, and expand the international visibility and competitiveness of Taiwan’s AI industry as a whole.
“Taiwan has mastered the three most important elements of the AI era: data, cutting-edge algorithms, and computing power (semiconductors). With a solid supply chain in Taiwan, we believe our prospects look promising for success in the AI era. I look forward to the day we can give back to our hometown instead of investing in businesses and national policies in other countries.”