[Interview] "We need to explore opportunities with AI semiconductors that utilize IP specialized in fabless industries"

Interview with Lee Seo-kyu, president of the Korea Fabless Industry Association.
"In order to strengthen industrial competitiveness, 'K-Fabless Growth Support Project' is required"
"It will help AI semiconductor development, foundry, and co-prosperity with academia"

[Sisa Journal e=Reporter Lee Ho-gil] "TSMC was able to become the No. 1 global foundry company because it has secured a number of proven semiconductor design assets (IPs) from leading fabless companies." It is impossible to develop the system semiconductor industry without the growth of fabless industry. "We will promote 'K-Fabless Growth Support Project' that supports R&D, prototype development, and leading technology research to secure competitiveness in system semiconductors."

Lee Seo-kyu, president of the Korea Fabless Industry Association (CEO of Pixel Plus Co.,Ltd.), emphasized that the fabless industry should be developed to develop system semiconductor capabilities, which account for more than 70% of the global semiconductor market. To this end, he stressed that training fabless design personnel and creating an ecosystem are key.

Samsung Electronics, along with TSMC, predicted that if support for fabless industries is provided, synergy between design and manufacturing could be created in the system semiconductor field.

He proposed a K-fabless growth support project as a way to strengthen industrial competitiveness. This is a concept that simultaneously promotes support for multi-project wafers (MPW) for system-on-chip (SoC) platforms and prototyping, and a project to strengthen universities' actual R&D capabilities. Through this, IP construction, commercialization, and next-generation technology development can be carried out at the same time.

The Fabless Industrial Association was launched last month to overcome the limitations that the center of gravity of domestic semiconductor policy has tilted to the manufacturing base. It is comprised of 105 fabless companies and will serve as a channel for delivering industry voices to the National Assembly and the government.

In his inaugural speech last month, He presented six tasks: ▲ Training design professionals ▲ Increasing private companies' participation in R&D projects ▲ Expansion of exchanges between finished product companies and fabless companies ▲ Promotion of fabless companies by activating mergers and acquisitions ▲ Creating healthy industrial ecosystems. 

--------- Q&A----------

- The Fabless Industrial Association's inaugural meeting was held last month. Did you start your activities?

Administrative procedures are still in progress, and the registration of the Ministry of Industry for activities will be completed within this month. The association's office has been set up at a shared office in Pangyo, and members of the association will start funding in general to secure office rent and labor costs for hiring full-time employees. A recent board meeting of the association was held to form a consensus on the direction of operation. I think I can start full-fledged activities from next year.


- It has announced the six major tasks of the association. What do you think is the most important task?

It is a training of human resources. The overall semiconductor industry has a serious manpower shortage, and most fabless companies, which are mostly mid-sized and small-sized companies, are dying. Technology development and commercialization are possible only when excellent human resources are present, but the number of design professionals produced a year is far from 100 people. The government has recently announced plans to foster semiconductor talent in various ways, but it is difficult to respond to the job shortage right now because it takes three to four years for training and manpower to come out.


- How can we recruit design professionals?

Using KAIST's Center for Semiconductor Design Education (IDEC) as a model, it is necessary to secure manpower through the 'micro-digris' (short-term completion course). Recently, IDEC recruited 40 students, and about 320 students applied. It can be seen that the demand for semiconductor education is quite high. Intermediate, advanced, and advanced courses are divided into four-month units, so even those who are not majoring in electronic engineering, such as those who majored in bio or chemical engineering, can take lectures. Fusion engineering is also possible, which requires activation.

I think it is necessary to create a system semiconductor academy that can develop experts in semiconductor design at a national level by utilizing this model. Apart from mid- to long-term measures such as expanding the semiconductor contract department, it is considered a way to train fabless personnel in a short time.


- Korea's share in the global fabless market is around 1%. What should I do to raise my teeth?

The government's national projects for R&D should reflect a lot of industry voices. In order for large fabless companies to emerge in Korea, a virtuous cycle structure is needed to develop technologies, make products, raise sales, and inject taxes paid into R&D support, but now the hurdle for national projects is too high. R&D should also focus on whether it can be mass-produced and commercialized, but now it tends to emphasize only a high technical level, not this.

The national project should be carried out in a way that contributes to the sales of companies, but there are few cases that have actually happened like this way. R&D and mass production are completely different dimensions. In the research process, it should focus on the possibility of mass production and change to a structure in which private companies can actively participate.


-There are many prominent companies such as Qualcomm, Nvidia, and MediaTec in the global market. Which fabless areas do you think the domestic industry should focus on to strengthen its competitiveness?

I think it's the car market. If it is converted to an autonomous vehicle, it will require a lot of power semiconductors, image sensors, and application processors (APs), so the point of differentiation can be the electronic market. I think that the effect of spreading to real life through robots and artificial intelligence (AI) will increase, and a very important market will be formed.


-Building an ecosystem along with individual companies' moves to explore new markets is likely to be an important task. The way to do this is...

A K-Fabless growth support project is needed. It is a huge opportunity that one of the foundry companies that can make advanced system semiconductors is a Korean company. I think it is necessary to take advantage of this opportunity to promote R&D support, prototype development support, and leading technology research support. We are proposing this to various organizations, and the response is positive. The primary project does not require that much budget, so I think it can start right away next year if you have a will.


- Please explain the proposed K-Fabless growth support project in detail.

It is to create a multi-purpose multi-project chip (MPC) that incorporates specialized technologies of domestic fabless companies into one chip by operating an R&D support business. Each company must have different areas of excellence in data processing, 3D image implementation, and security, and this function can be put into one chip. South Korean foundry companies have fine process technology of 14 nanometers or less, so they can be manufactured when they have infrastructure. New opportunities can be opened by creating AI-based semiconductors using fabless companies' specialized IPs.

The prototype development support project is to activate MPW. Now, only foundry companies have MPWs, so it takes a lot of time to develop them, and fabless companies produce MPWs to make digital products once every two quarters and analog products once every three quarters. In this way, many IPs can be developed, and it helps to win-win fabless and foundry.

The leading technology research support project is to develop the research capabilities of universities from a mid- to long-term perspective. When you ask the school what the most difficult part of the research process is, it is said that the wafer process takes six to eight months from design to product release. I think that if joint research between companies and universities is supported and MPW is activated to reduce the process period to two to three months, the industry and academia will be able to coexist.


- What do you think is important about the government's fabless support policy?

Support measures for revitalizing M&A are needed. In order to strengthen global competitiveness, it is necessary to increase the weight class through fabless company M&A, but in fact, it is not an easy problem because each company has different interests. It is not easy to find a solution. For this reason, I think it is necessary for the government to seek ways to revitalize M&A using parent funds in terms of fostering the industry.