South Korea’s Defense Dilemma: Why Secret Combat Data Is Stalling Innovation

Daniel Kim 2026.05.31 Views

Translation result

동행미디어 In military affairs, data can mean life or death. South Korea’s Ministry of National Defense and the National Intelligence Service must open access to “combat data” if innovative firms are to emerge and scale. (Cho Sang-geun, research professor, KAIST)

I hope government and industry will not only upgrade technologies but also collaborate to modernize institutional frameworks. (Jang Hyun-ho, head of the Advanced Technology Contracting Team, Defense Acquisition Program Administration)

On the 16th, following the forum “Era Forum: A New Age of War — The Present and Future of K-Defense,” hosted by Donghaeng Media Sidae at the Korea Chamber of Commerce and Industry in Jung-gu, Seoul, participants laid out priorities for K-Defense innovation.

Moderator Lee Jeong-dong from Seoul National University’s Graduate School of Engineering led a deliberative session that included Lee Yong-kwan, CEO of Bluepoint Partners; Jang Hyun-ho of the Defense Acquisition Program Administration; Cho Kyu-tae, director of the LIG DnA AI Lab; Cho Sang-geun of KAIST; Hong Sun-geun, chairman of Sidae; Lee Sang-eon, director of Sidae’s Institutional Innovation Institute; and Lee Moo-young, deputy director of the same institute.

Speakers urged that real-world combat data held by the Defense Ministry and the National Intelligence Service be shared with the private sector to foster startups with AI and other advanced capabilities and to strengthen national defense. While South Korea has a manufacturing base that can produce everything from raw materials to finished systems, experts said the defense sector lacks the combat data needed to drive rapid innovation.

Cho Sang-geun warned that access to real combat data can change mission timelines and strike accuracy by an order of magnitude. He pointed to the U.S. and Ukraine, where governments maintain extensive combat video databases and provide them to industry to accelerate weapon-system upgrades.

Participants criticized South Korea’s heavy emphasis on security for preventing startups from validating their technologies. Lee Jeong-dong said that if the National Intelligence Service or the Defense Ministry keeps data locked for security reasons, startups cannot even attempt new solutions. He called for a proactive approach to data sharing.

Below are excerpts from participants’ remarks.

동행미디어

You can’t develop drones without battlefield data

▶ Hong Sun-geun, chairman of Sidae (hereafter Hong)= I hear some foreign buyers purchase defense products made by Korean startups before the domestic market does. Why does that happen?

▶ Cho Sang-geun, research professor at KAIST (hereafter Cho)= The U.S. separates the channels that adopt startup technology: integrating tech into existing platforms, developing entirely new weapons, and funding long-term research. South Korea tends not to distinguish among these paths. An AI startup focused on artillery command systems went abroad first for that reason. It failed to account for domestic political and security dynamics. The same issue explains why some companies working on kamikaze drones in counter-drone systems haven’t received attention. To intercept a target drone, an interceptor generally must travel at least 1.5 times faster than the target. We still lack the capability to reach speeds around 300 km/h. Problems range from misdefined engagement envelopes to gaps in technical maturity.

▶ Hong= We should broaden the conversation on how acquisition processes can both nurture companies and strengthen defense. Even if interceptor drones can’t immediately reach 300 km/h, the government should back firms that show commitment. We need mechanisms that let companies iterate and improve after early failures.

▶ Cho= Exactly. Raising drone speeds requires engineering data — especially combat data. High top speed means little without realistic operational data. The state must address this. The National Intelligence Service and Defense Ministry should provide combat datasets. With solid combat data, domestic innovators could emerge and scale rapidly. Our planned dispatch of a Ukrainian combat-training analysis team was canceled, so we also missed access to foreign combat data. Historically, combat-training analysis has been treated in a bipartisan, nonpartisan way. That’s regrettable. Japan’s Self-Defense Forces, by contrast, have quickly absorbed real-world data through cooperation with the UK and Australia and are building mixed manned-unmanned combat systems. Military data truly affects survival — it’s not comparable to commercial economic data.

▶ Lee Jeong-dong, professor at Seoul National University’s Graduate School of Engineering (hereafter Lee)= We must release data that reflects new battlefield realities. The National Intelligence Service or Defense Ministry will likely resist on security grounds. If data remains locked, drone ventures cannot test new approaches. Companies demand data to survive, but I doubt the government will loosen controls without a clear framework.

▶ Lee Yong-kwan, CEO of Bluepoint Partners (hereafter Lee YK)= From a startup perspective, product development can proceed even without broad access to real combat data. Data and regulation are external variables; founders shouldn’t stake their companies solely on them. Advances in physics-based simulation now let teams generate large synthetic datasets from a small set of real examples. Simulations can mature technology close to a deployable state, even if they’re imperfect. Real combat data helps accelerate growth, but startups cannot depend on it alone.

▶ Cho= In modern warfare, the cycle for changing operational-security patterns has shortened to about two hours. Russian forces conceal armored vehicles beside homes and cover them with patterned tarps to evade AI reconnaissance. They change those camouflage patterns roughly every two hours. That’s why up-to-date combat data and real-time intelligence collection matter. Depending on access to real combat data, mission timelines and strike accuracy can differ by more than tenfold. The U.S. and Ukraine continually build and refine combat datasets and provide them to industry. Ukraine, in particular, has compiled an enormous database of combat footage. They can supply tailored datasets quickly for specific weapon-system development. By contrast, we overemphasize secrecy. Our intelligence and military organizations have faced internal disruptions, complicating core tasks like data sharing.

▶ Lee= This isn’t only the Defense Ministry’s problem. South Korean semiconductor firms often develop excellent technologies but only gain global recognition when a company like Intel adopts them. That dynamic puts global tech leadership in others’ hands and leaves us following. That pattern is a serious concern.

Shift from hardware-first to software-first

▶ Hong= South Korea’s defense industry still revolves around large conglomerates and hardware. Have any startups accelerated rapidly with support from the Defense Acquisition Program Administration? What types of startups should the sector encourage? Growing innovators would also raise the agency’s standing.

▶ Jang Hyun-ho, head of the Advanced Technology Contracting Team, Defense Acquisition Program Administration (hereafter Jang)= The idea of cultivating a Korean Palantir has emerged in defense circles. While details remain private, the government is exploring institutional measures. Officials have been listening to private-sector concerns and recognize the issues. We’re studying models like Anduril and Palantir and working across ministries to collect input and shape policy.

▶ Cho Kyu-tae, director of LIG DnA AI Lab (hereafter Cho KT)= The Defense Acquisition Program Administration and the Defense Ministry are making serious efforts, but industry still feels a gap in pace. Speed is critical when adopting software in defense. To keep up with AI-driven, software-centric change, we must reform procurement processes that remain hardware-first.

▶ Jang= A bill to simplify software acquisition procedures under the Defense Acquisition Act was proposed to the National Assembly in February. We need to upgrade institutional capabilities alongside technology. I hope public and private sectors can collaborate to address shortcomings step by step.

▶ Cho= The blue ocean for defense startups is personal combat systems — technologies that protect individual soldiers. Counter-drone systems, survivability equipment, and wearable tech that ensure soldier survivability remain underdeveloped. On the modern front lines, wired drone attacks using fiber links can defeat jammers; even U.S. forces have suffered significant losses. Gear that ensures individual survivability will become essential. Ukraine and the U.S. treat this area as a high priority and are investing heavily. On Ukrainian battlefields, transparent cloaks that defeat thermal cameras and wearable robots are already in use. Given Korea’s IT infrastructure and capabilities, we can be globally competitive.

▶ Lee YK= A U.S. defense drone company once planned to outsource production to China but chose Korea amid geopolitical tensions. That shows how highly Korean manufacturing ranks in global supply-chain realignment. If given clear objectives, Korea’s manufacturing base can execute them reliably.

▶ Cho= South Korea has a comprehensive manufacturing infrastructure that covers raw materials, components, and finished products. Provide the design, and we can rapidly mass-produce world-class hardware. The U.S. National Defense Strategy released in January highlights strengthening alliance-based manufacturing as a priority. With China pursuing a project to mass-produce one million drones, the U.S. aims to work with manufacturing-capable allies like Korea, Taiwan, and Japan to produce and field equipment locally rather than concentrate production in the U.S. If Korea secures that manufacturing-hub role, it would gain substantial diplomatic leverage.