[20 Years Ago in the IT Industry] The Conditions for a Software Powerhouse Were Embedded

Embedded software (SW) drew attention in the mid-2000s as a next-generation industry and was seen as a field with strong growth potential. Its application areas were also diverse, ranging from mobile phones, TVs, automobiles, factory equipment, and medical systems to aviation systems. In the global market, there was no dominant powerhouse, so it was viewed as an opportunity for Korea's domestic industry.

Korea had already secured competitiveness in industries that applied embedded SW, and the sectors cited as strengths included semiconductors, consumer electronics, mobile phones, and automobiles. With a solid network infrastructure as well, there was growing expectation that the country could emerge as a global embedded SW powerhouse if its hardware strength could be linked to SW.

Accordingly, the government designated embedded SW as a new growth engine and pushed for its development. Research institutions such as the Electronics and Telecommunications Research Institute (ETRI) also moved ahead with core technology development.

The embedded SW industry was showing a reality different from growth expectations. Dependence on overseas core technology was high, specialized talent was lacking, and although many companies were participating in the market, most were small. These companies relied more on contract development for large corporations than on their own products or licensing businesses. This newspaper took a close look at the market conditions, government policies, industry problems, and possible solutions.

The embedded SW market was worth KRW 8 trillion, and foreign-made products accounted for 85.8% of that total. Embedded SW refers to software installed in devices and equipment to perform specific functions, and it is distinct from general-purpose computer software. Its scope included OS, middleware, and applications, referring to the software area excluding hardware.

Embedded SW began with industrial equipment control. At the time, it was being used not only in consumer appliances such as mobile phones, TVs, and washing machines, but also in trains, airplanes, and elevators. Its application areas had expanded to factory and home automation, security and disaster prevention, education, personal handheld terminals, communications equipment, distribution and payments, transportation, healthcare, military and aviation, and the environment, and the range of use was rapidly widening.

As devices became more digital and multifunctional and network connectivity expanded, demand for embedded SW was also expected to rise quickly. Even amid that trend, the industrial reality remained far from expectations, with dependence on overseas technology, a shortage of workers, and a small-scale business structure.

Embedded SW is drawing attention as a high value-added industry because its output per worker is higher than in other SW fields. In the comparison of output per worker, package SW came in at KRW 60 million, digital content at KRW 130 million, IT services at KRW 180 million, and embedded SW at KRW 190 million, making embedded SW the highest among the categories compared. Park Tae-wan, a former IT official at the Ministry of Information and Communication, explained that software accounts for 20% to 30% of development costs in automobile and defense projects, meaning higher SW development costs can lead to higher product prices.

Analysts also said Korea's chances in the market stemmed from its hardware competitiveness. Digital TVs, smartphones, and IP set-top boxes were cited as products competitive in the global market, and the view was that success would be more likely if domestically developed embedded SW technology were combined with competitive hardware. Network infrastructure was also mentioned as a domestic strength. Kim Tae-young, CEO of Ensarch Korea, said that because population density is high around major Korean cities, it is easier to design and implement infrastructure for IT-related services, and that Korea is well-suited to serve as a test bed for various new IT technologies.

In addition to these conditions, the fact that domestic efforts in ubiquitous computing were already under way was also linked to Korea's advantage. Kim Heung-nam, head of the embedded SW division at ETRI, said ubiquitous computing implementation was already being attempted in Korea, and based on that, expressed the view that Korea could move ahead of overseas competitors.

As the operating system market changed and embedded devices became more complex in function, the early market structure centered on dedicated RTOS for industrial and military equipment began to give way to broader influence from general-purpose environments such as Windows CE and embedded Linux. VxWorks, pSOS, VRTX, and QNX were cited as products that shaped the early market. Kim Heung-nam said that while specific company products were used at first, Linux had recently become dominant because of compatibility.

Along with those changes, development of embedded Linux solutions accelerated, backed by the advantages of open technology. Development methods also shifted, with platform-based approaches spreading in which needed middleware and application SW were selected and installed.

An industry survey conducted by the Embedded Software Council of 179 demand companies and 134 supply companies found that Korea's embedded SW market was worth about KRW 8.1 trillion in 2005, accounting for 7% of the global market. By sector, industrial automation was the largest market, followed by information appliances and wireless communications.

Separate from market size, foreign products accounted for a high share in core SW and foundational technology areas. With the domestic share of core SW being low, foreign license products accounted for 85.8% and domestic license products for 12.9%. In the development tools market, foreign products accounted for 95.8%, while domestic and foreign shares in application SW were found to be the same.

As this structure of heavy dependence on overseas technologies for foundational areas such as OS and development tools became clear, concerns also arose. There were worries that deeper dependence on foreign technology could restrict the growth of domestic SW companies, and that a weak domestic industrial base could lead to a renewed rise in reliance on foreign products.

These trends were also linked to concerns about a possible vicious cycle of foreign dependence and industrial fragility. Another concern was that even if Korea had the potential to produce globally competitive finished products and hardware, growth in the domestic SW industry would remain limited if the core SW inside finished products depended on overseas technology.

At the same time, stark polarization within the industry became evident. Companies with fewer than 300 employees made up 99% of firms, but their share of sales was only 1.9%. By contrast, firms with 300 or more employees accounted for just 1% of companies, yet they represented 98.1% of total sales among companies with 300 or more employees, showing that sales were concentrated in a few large firms.

The main areas of focus for domestic small and venture embedded SW companies were games, mobile phone AV, consumer electronics, and wireless communications. Industry participants called for moving away from an application SW-only focus and securing capabilities in OS and middleware, while also expanding from mobile and consumer electronics markets into automobiles, ships, aviation, and defense.

At the same time, many domestic embedded SW companies generated revenue mainly through contract and commissioned development based on orders from large corporations rather than through their own product licensing sales. A survey at the time showed that the contract development market was five times larger than the licensing market.

An industry source said that most domestic companies remained focused on contract work rather than entering the licensing market. The source also said the revenue structure remained unstable because many contract projects ended as one-off deals.

Another source pointed to the fact that many companies participating in international embedded exhibitions were small in scale. Based on that, the source said domestic companies lacked both capital and manpower.

As a solution, Kim Heung-nam emphasized a shift in business models and said domestic companies needed to build royalty-based business models. He said such models were necessary to increase solution value added, and that differentiated solution development and competitiveness were prerequisites for that shift.

The government viewed embedded SW as a high value-added industry and pushed a development policy centered on talent training and technology development. The industry agreed with the policy direction but pointed to the lack of short-term support and specificity, and there were also criticisms that few companies had actually received government support. One company source said that even after checking support programs, they gave up applying because communication with the person in charge was poor. In this situation, observers noted a gap between declarations of industry promotion and companies' sense of policy effectiveness.

Questions were also raised about the connection between government R&D and the industrial front line. The industry pointed out that even when companies received government-developed technologies, follow-up development stopped after the project ended, and said version upgrades and additional development had to be funded by the companies themselves. There were also assessments of insufficient support for tax incentives tied to R&D, insufficient support for overseas expansion, and insufficient follow-up support related to national certification and standardization.

The industry raised additional issues beyond expanding subsidies, saying that financial support alone was not enough. It also called for systematic research into overseas SW policies, a reduction in domestic developers being drawn into unclear copyright disputes, and a stronger institutional foundation to protect the intellectual property rights of domestic technologies.

For the industry-academia-research consultative body, there were calls to move beyond meetings and come up with concrete measures for industrial development. The industry said the consultative body needed to prepare practical action plans.

In response, the government set up an embedded SW technical support center in Daegu, centered on government policy, technical support, and talent training. It pushed for technology transfer and commercialization support for ETRI's development platform 'Qplus,' and the scope of government technology transfer was set from product development through maintenance. It also promoted joint development of mobile convergence terminals and efforts to build industry-academia-government-research networks.

The project was operated based on Daegu's concentration of electronics and auto parts manufacturers. Plans were also outlined to expand the effort to other regions if the project proved successful.

As it promoted technology diffusion and commercialization, the government applied domestically developed technology examples such as ETRI's embedded Linux-based OS 'Qplus' and ETRI's development tool 'Esto' to pilot projects for Jeju telematics and the National Robot. In addition, school-level education was conducted based on 'Skill Set,' which defined the required technology levels for mobile communications, set-top boxes, telematics terminals, PDAs, ITSoC, and industrial control fields.

In response, Kim Ki-cheol, executive director of the Embedded Software Council, was quoted as saying that government policy had improved awareness of the industry. He pointed to the increasing number of participants in embedded SW competitions each year and the upward trend in related academic departments as evidence.

The government also pursued policies to secure advanced talent. The Information and Telecommunications Research Centers (ITRC) provided opportunities for master's and doctoral students to take part in major industry-academia joint research projects, while the ITRC at Pohang University of Science and Technology focused on mobile, the ITRC at Konkuk University on Linux, and the ITRC at Sun Moon University on development environments. The number of graduates from ITRC programs was 43 in 2003, 62 in 2004, and 49 in 2005, for a total of 156, and the government also supported practical talent training linked with educational institutions such as Bitcomputer.

Potential talent in embedded SW was identified through competitions and through in-house training and contests at companies. The embedded SW competition was used as a channel to discover potential talent, and the number of entries rose from 74 in 2003 to 204 in 2005. Kim Heung-nam said there was an example from the previous year's contest in which LG offered a job to a prize winner, and he added that there were many cases of prize winners actually being hired by large companies. Alongside this, the industry worked to secure its own workforce.

DigitalCube, a Linux-based PMP company, held an open-source contest to improve technology and train personnel. MDS Technologies, an embedded SW specialist, operated MDS Academy, an in-house professional training institute. In this way, companies moved to secure talent and strengthen capabilities not only through competitions but also through contests and training organizations.

At the same time, companies, the government, and research institutions jointly pursued technology development and institutional reform. The government supported small and medium-sized companies in developing their own technologies, covered part of the cost of prototype production, and added an embedded SW category to the New Software Product Award. ETRI prepared follow-up technology development for mobile convergence based on embedded Linux, and also pushed support for securing intellectual property through legal advice, search, and analysis, as well as the creation of field-specific standards and certification systems. It also responded to international standardization movements such as OMA for mobile application services and CELF for embedded Linux standardization in consumer electronics.

Competition in embedded SW unfolded at the national level, not just the corporate level. The United States invested about USD 200 million in embedded SW R&D through the National Science Foundation (NSF) and the Defense Advanced Research Projects Agency (DARPA), and the U.S. National Science and Technology Council (NSTC) designated embedded SW as a major R&D field for the 21st century. In Europe, EUREKA's ITEA invested 3.2 billion euros from 1999 to 2007 in embedded SW development for vehicles, transportation, consumer electronics, and office systems, while IST invested 540 million euros from 2003 to 2006 in network systems, next-generation DSP, adaptive system SW, and more.

Japan promoted R&D around TRON (The Real-time Operating system Nucleus) and also supported the spread of standard RTOS. Kim Ki-cheol explained that although TRON was privately led, it was in practice accompanied by government support, and said TRON was widely embedded in Japanese consumer electronics, meaning domestic consumers could use TRON-based products without even realizing it.

This case showed that Korea's challenge at the time was a separate issue: having a strong finished-product industry and nurturing a domestic SW ecosystem within those products. The fundamental obstacle to industrial growth was the ecosystem.

The structure was shaped by large companies leading development and small SW firms relying on subcontracting and commissioned work. As a result, specialized firms struggled to accumulate their own technologies and products, and behind this was the relatively better treatment offered by large companies. Accordingly, more advanced talent moved to large firms, while small companies concentrated on the difficulties of technology development and talent recruitment.

Another problem was the difficulty of accurately grasping the market itself. The reason cited was the intermediate-goods nature of embedded SW, which is fed into finished products. As a result, it became difficult to calculate independent market size and trends, and there were also criticisms of the limitations of major overseas market research firms' surveys at the time.

In this process, some industry participants were also confused about whether their own business fell under embedded SW, creating uncertainty. Ultimately, this led to a lack of market definition needed to establish policies for revitalizing the industry and a lack of basic statistics needed for those policies.

The small scale of the industry led to inefficiencies in technology development, and small companies had difficulty understanding what other firms were developing and where the market was heading. As a result, cases of duplicate development of similar technologies occurred.

The industry pointed out that joint research and collaboration among small firms were needed to reduce wasted development costs. Seo Young-jin, president of Miji Research, also said small companies had limits in tracking technology development and market trends, and stressed collaboration among companies as the solution.

However, there were limits to how far government policy could solve these problems. Kim Ki-cheol said government policy stayed at the level of supporting product sales channels and made it difficult to promote technology transactions or M&As between companies, and therefore the industry needed a separate structure for sharing technology and market information and for cooperation among companies.

At the same time, intellectual property issues came to the fore between large and small companies. Critics said small companies were not properly recognized for their own intellectual property when carrying out contract development work for large firms, and concerns were also raised about large companies' distrust of the technology capabilities of domestic small firms. As a result, large companies tended to prefer foreign products over domestic ones.

One proposed alternative was pilot projects. The logic was that large companies could reduce adoption risk by trialing small-firm products before full-scale commercial projects, while small firms could gain opportunities to prove their technology through pilot deployments.

Kim Heung-nam said that Qplus, which they developed, had been applied to the second-stage pilot project for Jeju telematics and had received better evaluations than foreign products. He said such pilot projects could be a good opportunity to assess domestic technological capabilities. However, he also said that for domestic products to be selected in the real market, small companies first had to secure competitiveness in both technology and products.

It was also pointed out as a problem that companies concentrated too heavily on fashionable fields, and there were calls to avoid following short-term market trends. The need to train specialists across diverse fields was also emphasized, along with the idea that new markets should be created on that basis.

In the embedded SW industry, dependence on foreign foundational technologies, the small scale of specialized firms, a business structure centered on one-off contracts, a shortage of advanced talent, and a lack of cooperation between large and small companies continued, and these factors were interconnected. As a result, it became harder to accumulate proprietary products and intellectual property, which in turn made stable revenue and talent acquisition difficult.

Even with support for technology development, weak industrial foundations for connecting support results to ongoing business were identified as a problem. For that reason, there were limits to continuing business based on the results of support programs.

Achieving embedded SW power-house status was not possible through the development of a single domestic product alone, and raising localization rates alone was not enough. The key task was to build an industrial structure that linked the capabilities of the hardware industry to in-house SW technology and the growth of specialized firms.

At the same time, the role of embedded SW was expanding. The target fields were physical products such as automobiles, robots, home appliances, and industrial equipment. Its existing role was device control, but its development direction was moving toward the direct execution of AI models. The section topic ran from embedded SW to on-device AI.

ETRI presented an analysis of 2025 on-device AI technology trends and cited real-time performance, energy efficiency, privacy, and independence as key technical requirements. In an analysis of system SW restructuring, it summarized that the existing embedded SW-centered structure was shifting toward an AI-centered structure.

ETRI said the background to this shift lay in the need to use a device's own resources and power efficiently and to run AI models stably, and analyzed that the role of system SW had therefore expanded to operating systems, compilers, and AI execution frameworks. In particular, because automotive and robot systems require both real-time performance and high reliability, it said SW capabilities in those systems have become as important as hardware performance.

ETRI also presented links between specialized fields of accumulated research and changing application areas. ETRI pushed to expand the focus of its embedded SW research in the 2010s, with expansion targeting defense, aviation, and automobiles. These fields require high reliability and stability, and a representative output was the real-time operating system 'Qplus-AIR' for unmanned aircraft.

Qplus-AIR secured compliance with international standards and top-level safety by meeting ARINC 653 requirements and obtaining DO-178B Level A certification. It then went through integrated testing of ETRI's flight control SW, and high-reliability verification was achieved through that integrated testing.

This trend is presented as an example of research accumulation in system SW expanding into specialized fields such as aviation and defense.

Meanwhile, the issue is becoming especially pronounced in the automobile industry. As SW implementation of vehicle functions and services, along with continuous improvement through SW, expands, the shift to software-defined vehicles (SDV) is under way, and the share of SW in the competitiveness of automakers and parts suppliers is also growing.

According to SPRi's 2026 analysis of Korea's SDV industry, the domestic SDV sector was assessed as having entered a stage of expansion in terms of R&D and manpower input, and the number of workers related to SDV also increased. However, the report also identified an early stage in terms of SW-centered internalization, and it confirmed a low share of core SW personnel and a lack of connection between visible investment and structural change within the industry.

SPRi said that to achieve the SDV transition, cooperation between automakers and SW and ICT companies needs to be strengthened. For that, it said the industry must move away from simple outsourced SW procurement relationships and form a cooperative structure capable of joint development and technology accumulation, while also securing SW personnel that match industry demand. SPRi identified the formation of a cooperative structure and securing SW talent as the core of competitiveness.

SPRi noted that in the past, embedded SW was technology that added functions and value to competitive hardware. But now, it said, the status of SW has changed to form the basis for defining the functions of automobiles, robots, home appliances, and industrial equipment, and its scope has expanded as the core foundation for running AI. It said the spread of on-device AI is making system SW itself a key area of competition.

SPRi said that a strong manufacturing sector does not necessarily mean a strong software industry. The key, it said, is whether core SW can accumulate in-house technology and talent, and whether specialized firms can sustain commercialization and foster intellectual property. This awareness connects with the question raised 20 years ago about the path to becoming an embedded SW powerhouse, and the name for that continuum today is 'AI internalization.'

Source: IT DAILY · Kim Byung-joo
Original: https://www.itdaily.kr/news/articleView.html?idxno=241057

References

• https://www.itdaily.kr/news/articleView.html?idxno=241057