ASICLAND’s Comprehensive Design Services

In Korea’s domestic market, increasing saturation has made it difficult to gain new market share. Experts now suggest that this could be the ideal time for Korean companies to consider exporting, especially with regulations like the Inflation Reduction Act and the CHIPS Act promoting onshoring and supply chain restructuring—benefiting Korean multinationals. Do you agree that this is a favorable time for small and medium-sized enterprises to diversify and internationalize? What do you see as the primary challenges and opportunities for these SMEs when operating abroad?

With regard to the global supply chain, while current trends don’t necessarily offer direct advantages for Korean multinationals, they do present significant benefits for Korean fabless companies. The competition among the US, China, and Taiwan is intensifying, and, while competing with industry leaders in the US, China, and Taiwan, we must focus on price competitiveness and technological innovation. Now that China is increasingly excluded from US-led supply chains, this opens up new, substantial opportunities for domestic fabless companies.

The COVID-19 pandemic has also spurred demand for 5G, electric vehicles (EV), and AI, presenting further openings for the Korean semiconductor and fabless market. However, I believe that Korean fabless companies still have room to improve in maximizing their strengths compared to large corporations like Samsung Electronics and SK Hynix. Although there is a significant opportunity before us, we must continue to heavily invest in talent and R&D. With strong support from the Korean government, I’m optimistic that a few Korean fabless companies will emerge as “hidden champions” in the near future.

In terms of strengths, Korea leads in memory and display sectors. However, we are still not dominant in system semiconductors, though we see potential to grow in areas like NAND access control and DRAM-Less solutions. The unilateral export restrictions from Japan in 2019 underscored the importance of internalizing production capabilities. This need for self-sufficiency, coupled with the broader AI trend, is driving increased investment in these fields.

You mentioned the relative weakness of Korean fabless firms in the system semiconductor market. This is an urgent priority not only for companies but also for the government. How should Korean semiconductor fabless firms work with the government and other industry players to strengthen their market position? Which specific semiconductor sectors should be prioritized?

The Korean government has provided substantial policy support for R&D, human resources, and tax incentives, but I believe tax credits here still fall short compared to Taiwan, China, or the US, limiting their impact for Korean fabless companies. The inconsistency of R&D policy support across different administrations has also been challenging—when companies miss a policy window, they often cannot benefit significantly from R&D support. We need more continuous, predictable R&D backing from the government.

Looking at Korea’s macroeconomic context, we saw some R&D support in the early 2000s, but this decreased for the semiconductor industry in the 2010s. Fortunately, since 2020, the government has reinvigorated its R&D support for semiconductors. For long-term growth, we should focus not only on AI innovation but also on areas where Korea has core strengths. We must address gaps in fields like power semiconductors, system semiconductors, and EV technology, where investment has been lacking despite solid infrastructure.

You mentioned the growing role of the AI megatrend. This is also evident in your company, as we've seen your sales ratio for 12nm to 7nm products increase significantly over the past year, driven by edge AI applications. Traditionally, semiconductor revenue has come from legacy applications such as automotive and consumer electronics. But today, with emerging technologies like AI, new systems like HBM are reshaping the market. Looking ahead three to five years, what new technologies or applications do you believe will be the primary growth drivers for your company?

You’re exactly right. In the custom HBM field, Korea is not yet a dominant player in the system semiconductors used in custom HBM, which large conglomerates still largely control. TSMC is also turning its focus to custom HBM due to its significant market potential. Logic and custom HBM are set to become the next generation’s high-growth markets, and I believe SK Hynix will likely be a major beneficiary in the custom HBM sector, along with companies like us, which operate as design houses. Another key area of focus for us is CXL, which memory semiconductor companies need to optimize AI functionality. In our development phase, we’re concentrating on custom HBM, CXL, and HPC—these three pillars will be the main drivers of growth for our company. Additionally, we are launching mass production of our edge AI accelerators and edge AI computing solutions, which we have already developed.

Founded in 2016, Asicland quickly emerged as a prominent ASIC provider, collaborating directly with TSMC. By 2018, the company had become an approved ARM design partner and had successfully designed 5nm chips. What do you see as the key factors behind this impressive growth and success?

I’m often asked this question, and I’d attribute much of our success to a combination of luck and strategic decisions. From the outset, our goal was to go beyond backend design as a design house and to expand into the frontend, utilizing ARM technology to work on CPU design. Becoming a TSMC VCA was perhaps the most significant milestone that fueled our growth. From the start, we aimed to partner with TSMC, a goal we initially thought would take around seven years to achieve. However, we were recognized much earlier because TSMC’s previous Korean partner signed an exclusive contract with Samsung Electronics, creating an opening. Thankfully, we were able to step in as TSMC’s new partner and expand into frontend design within just two to three years of our founding.

It’s clear that collaborating with TSMC is a valuable asset for your company. As a member of TSMC’s Value Chain Alliance (VCA) program, you help extend TSMC’s reach by offering direct design services as an independent company closely aligned with the foundry. While Taiwanese firms like ALCHIP and GUC are major fabless players and companies like IMEC.IC-link and TOPPAN TDC belong to large international groups such as IMEC in Belgium and TOPPAN in Japan, Asicland holds a unique position as an independent and rapidly growing SME. What specific advantages does Asicland offer compared to other VCA members and design houses?

Companies like Toppan and IMEC, primarily based in Europe, could certainly present a competitive edge in the Korean market. However, their main challenge lies in language barriers and a lack of on-site support, which is crucial given the sophistication and rapid evolution of Korean clients. Our ability to provide immediate on-site support sets us apart and is invaluable to Korean clients who require responsive and adaptable services.

Another key advantage is our range of platforms, such as our data center, AI accelerator, and IoT platforms. The biggest competitive edge of these platforms is their ability to drastically reduce the development timeline, which in turn lowers manufacturing costs. Though our initial investment in these platforms required substantial capital, this approach has attracted more clients and enabled us to enter the US market. To support high-performance computing (HPC), we’ve also acquired a 10 billion KRW Neoverse CSS license.

For instance, we supply edge computing platform templates to AI companies and customize them based on client needs. By significantly shortening the development timeline—from two years to as little as eight months—we pass those time savings on as a tangible benefit for our clients.

Since you mentioned the US market, this has clearly been an important year for your company due to the deal you signed with an American firm. This appears to be your first major contract in the US. I find it intriguing that Asicland provides turnkey services—from spec-in to testing. Among North American fabless companies, most Level 2 and Level 3 firms typically handle specific parts, such as backend and OSAT, or frontend and OSAT, but few independent companies offer such comprehensive end-to-end services. Do you see this as a market gap or an opportunity?

Mid-level companies, such as startups, often need frontend design but typically only find design houses that specialize in backend. Our technological expertise and ready-made templates allowed us to enter the US market by collaborating with PHYTunes Inc., a startup searching for a design house that could handle analog frontend design. We saw an opportunity to meet this need.

PHYTunes may have initially considered partners in India or other US-based design houses, but there may have been communication challenges or cost discrepancies with those options. With our templates and technical proficiency, we could provide them with the cohesive support they needed.

With PHYTunes, you’re working on their new multiband RFIC. Based on this deal, how do you envision the evolution of your operations in the North American market? How significant will the U.S. market become for your business over the next year?

Mass production for PHYTunes will take some time; we view this as a mid- to long-term plan since it’s still in the early stages. We’re optimistic about PHYTunes' success, and we hope to share in that growth. Our goal now is to establish ourselves as a successful analog frontend design service provider in the U.S. The PHYTunes contract serves as a bridge to promote our digital templates in the American market. We’re currently in active discussions with several prospective clients, both large and small, and while outcomes aren’t certain yet, these conversations are promising.

We’ve discussed edge AI chips and designs several times. Edge design semiconductors have a wide range of applications, from image recognition to autonomous driving, where chip designs may vary. While much of the semiconductor race focuses on achieving the smallest nanometer scale and highest performance, edge AI semiconductors are more about balancing power and efficiency. How do you guide your customers in selecting the optimal design for their chips, balancing power and efficiency? Could you share a success story?

As an SoC operator, we employ multiple schemes to supply low power solutions to our clients. Since each client’s application is unique, they typically pilot-test our schemes on-site with their MPUs. We integrate key low-power strategies, such as multi-VDD and power gating, allowing clients to configure systems to suit their needs. We provide guidance on the default power schemes we’ve implemented, and from there, clients determine whether low-power mode is suitable for their MPUs. This approach has received very positive feedback, and our clients have been highly satisfied with the results of our power schemes.

In Q3 2024, you opened a new R&D center in Taiwan. How will this new center help you attract more clients, enhance expertise, and draw more talent to your company?

The primary purpose of establishing our R&D center in Taiwan was to advance our work in 3-nanometer technology and to support 2.5D and 3D packaging development. Currently, Korea lacks engineers specializing in 3-nanometer nodes or 2.5D and 3D packaging, even within major companies like Samsung and SK Hynix. However, Taiwan, particularly through TSMC, has a skilled workforce in these areas. Our aim with this Taiwanese R&D center is to cover all aspects of 3-nanometer chiplets and 3D packaging, positioning us to attract more AI and HPC companies, along with global MNCs such as SK Hynix.

Let’s imagine we have another interview with you in 2026, which will mark Asicland’s 10th anniversary. What achievements would you like to share in that future interview?

By 2027, my goal is to initiate mass production of edge computing AI solutions and to advance development in CXL and HPC, all at the 5-nanometer scale, by the end of 2026 or early 2027. This progress would be key to reaching our revenue target of 200 billion KRW. In two years, I also hope we will have firmly established our presence in the U.S. market.

For more information please visit: https://www.asicland.com/