Centered on unicorns—unlisted companies valued at 1 trillion won or more—Korea's startup ecosystem is rapidly reorganizing around technology. While platform and app-based services once powered growth, now deep-tech startups in AI, semiconductors, and robots are emerging as the axis of growth, touting research and development (R&D) as their core competitiveness.
Deep-tech startups staff 70% to 90% of their workforce with R&D personnel and are focusing investment on core technology development and securing top talent rather than marketing. As technological prowess in the AI era becomes the key factor that determines corporate competitiveness and corporate value, the emphasis is shifting from short-term results to upfront investment for future competitiveness.
◇ Deep-tech strategy to grow unicorns: "Investment in technology over sales"
A representative case is Rebellions, an AI Semiconductor design unicorn. Rebellions staffs more than 75% of its roughly 350 employees in R&D, with master's and doctoral-level researchers accounting for more than half of them.
Rebellions is continuing R&D investment on a scale that far exceeds sales as it prepares to mass-produce next-generation AI Semiconductors. Rebellions recorded more than 30 billion won in sales last year. It is difficult to evaluate the competitiveness of growth-stage deep-tech startups by the "R&D investment ratio to sales" as with general manufacturing companies. Because technological prowess is the core asset that determines corporate value, the structure prioritizes upfront investment for securing core technology and top research talent over short-term revenue.
The company is also aggressive in securing key talent. Rebellions last year hired Kim Hong-seok, former head of R&D at Google Korea, as head software architect. A Rebellions official said, "We are focusing our capabilities on securing technology competitiveness at a global level by successively recruiting researchers from global big tech, including Google, Nvidia, Meta, and Intel."
Upstage, an AI model development unicorn, is also pursuing an R&D-centered growth strategy. Founded in 2020, Upstage is carrying out the Ministry of Science and ICT's independent AI foundation model project. The fact that a startup just over five years old is undertaking a national AI research and development project is cited as a representative example showcasing the technological prowess of Korea's deep-tech startups.
More than 70% of Upstage's roughly 190 employees are developers. This year, it expanded global joint research by bringing Stanford University Professor Choi Yejin and New York University Professor Cho Kyunghyun—world-renowned scholars in Deep Learning—into the foundation model project. Founder Kim Seong-hoon, the CEO, also previously served as a professor at the Hong Kong University of Science and Technology and has built an R&D-centered organization since the company's early days.
Holiday Robotics, a Humanoid Robot startup, is also accelerating the expansion of its R&D organization. Recognized with a 1 trillion won valuation after raising about 150 billion won in its recent Series A, Holiday Robotics has 42 developers, or 87.5% of its total staff. The organization is centered on researchers from the Massachusetts Institute of Technology (MIT), Seoul National University, KAIST, and Samsung Electronics, and it plans to expand its R&D headcount to more than 100.
◇ Surge in deep-tech investment, and concerns over an "AI tilt"
This R&D-centered growth strategy also aligns with the agile organizational culture unique to startups. Large corporations often go through multiple layers of decision-making in the R&D process, but startups can have founders or the chief executive officer (CEO) communicate directly with the R&D organization and quickly decide research directions. In industries like AI and robots, where technology must be advanced through repeated failures and improvements, such agility serves as an important competitive edge.
Choi Byeong-cheol, president of the Korea Venture Startup Academic Society (professor of business administration at Hankuk University of Foreign Studies), said, "A startup's greatest strength is speed," and added, "What differentiates them from large corporations is the ability to quickly implement ideas into technology, immediately reflect market feedback, and create a virtuous cycle of further advancement." He also said, "In the AI era, as technology and talent become increasingly important, R&D-centered growth strategies will be further strengthened."
Investment in deep-tech startups is also growing rapidly. According to the startup investment information platform The VC, investment in AI and Robotics in the first half of this year totaled 2.685 trillion won, a 5.8-fold increase from the same period last year (458.8 billion won). The analysis is that investors' funds are concentrating on companies that have secured core technology and research talent rather than short-term results.
The government is also expanding R&D support to foster deep-tech startups. The Ministry of SMEs and Startups this year reorganized the TIPS (TIPS) R&D program, which supports innovative startups, into a customized, stage-specific system such as scale-up and global. The support size was also significantly expanded from the previous 500 million to 1.2 billion won to a maximum of 2 billion to 6 billion won, and the related budget increased from 647 billion won last year to 1.0355 trillion won this year.
However, there is also caution in views of the deep-tech investment boom. Because AI, semiconductors, and robots require large-scale R&D funding over long periods and considerable time to commercialize, there is a possibility that businesses may fail to commercialize despite high technological capabilities.
Yoo Hyo-sang, head of the Unicorn Management and Economics Research Institute, said, "Deep tech can yield high revenue if successful, but it is also a field with a high risk of failure," and added, "If investment and talent become excessively concentrated in AI and robots, the growth base of other innovative industries could be weakened, so we also need to consider building a balanced ecosystem."