The state must lead the battle for R&D commercialization

Hong Sung-joo

The author is a senior research fellow at the Science and Technology Policy Institute.

The rules of the game have changed. For decades, the battlefield of technology belonged to corporations. Companies that made better products faster emerged victorious, and Korea’s semiconductor and display industries became symbols of that success. But that era is fading. The struggle has shifted beyond localized corporate competition into a full-scale contest between national systems. The outcome is now determined not simply by technological superiority, but by the capacity of states to dominate entire technology value chains.

From left, Rep. Jodey Arrington, Republican of Texas, Sen. Ted Cruz, Republican of Texas, Oracle executive Clay Magouyrk and Sam Altman, CEO of OpenAI, answer questions during a news conference on the Stargate AI infrastructure project in Abilene, Texas, on Sept. 23, 2025. [AP/YONHAP]

The United States is treating massive AI infrastructure projects such as Stargate as strategic national assets while using tariffs and subsidies to reorganize global supply chains around itself. Japan is rebuilding the entire process of advanced 2-nanometer semiconductor production through Rapidus, a public-private venture backed by the state. Europe, meanwhile, is also moving to secure self-sufficient regional supply chains. These are no longer simple industrial policies. They represent a shift toward a new techno-nationalist system that merges technological dominance with national security.

The decisive battleground in this competition lies in the dark space between laboratories and markets — the “Valley of Death.” It is the gap where long-term research and development (R&D) collide with the short-term logic of markets and investment returns. This is precisely where the state must step in. Only when governments are willing to shoulder risks that private firms avoid can the valley become a strategic front line for technological sovereignty.

The United States recognized this challenge decades ago. The Defense Advanced Research Projects Agency, or Darpa, was established in 1958 after the Sputnik shock. It later evolved into organizations such as ARPA-E in energy and ARPA-H in health care. These institutions deliberately operate in areas considered too risky for private investment yet too market-oriented for conventional public research institutes.

Their philosophy is straightforward. If every project succeeds, the government merely spent taxpayer money on safe ideas. Failure is not treated as a flaw, but as evidence that the system is genuinely pursuing innovation.

The strength of the U.S. model lies not simply in tolerance for failure, but in its ability to push technologies into the market. ARPA-E, for example, deploys “Technology-to-Market” teams from the earliest stages of research to help projects identify viable business models. Some projects supported by ARPA-E later attracted private investment several times larger than the original government funding, successfully crossing the Valley of Death.

Germany’s Sprind represents a European response to the same challenge. Created in 2019 amid concerns that Germany was producing knowledge while profits flowed overseas, Sprind operates as a government-owned limited company with the flexibility of a private enterprise. In 2023, Germany enacted the “Sprind Freedom Act,” limiting government oversight and guaranteeing institutional autonomy. The agency can distribute funding within weeks and even directly establish subsidiaries when necessary.

Korea, by contrast, remains detached from this structural competition. Although the country ranks among the world’s leaders in R&D spending as a share of GDP, its technology trade balance remains chronically in deficit. More than 70 R&D support organizations exist, yet few institutions are genuinely designed to absorb risk. High-risk projects still struggle to pass bureaucratic review processes, and innovation organizations capable of bridging the Valley of Death remain weak.

As Korea’s R&D system confronts these structural limitations, AI is emerging as a potential game changer, transforming innovation in two major ways. First, it dramatically compresses research productivity. Tasks such as discovering pharmaceutical candidates or predicting protein structures, once requiring years, can now be completed within months. In effect, AI is narrowing the Valley of Death itself.

Researchers at the Georgia Institute of Technology conduct quantum computing research supported by the Defense Advanced Research Projects Agency. [GEORGIA TECH]

Second, AI is becoming a strategic tool for innovation management. By analyzing vast collections of papers, patents and market data, AI can identify connections invisible to human researchers. It can support the judgment of program managers and improve the precision of national innovation strategies. This is why the United States increasingly views AI not merely as another industry, but as the intellectual framework for redesigning the state itself.

Korea, however, still tends to treat AI as a matter of industrial competitiveness alone. Debate remains centered on who can build better models rather than how AI should transform national governance and decision-making structures. The more urgent question is how AI can be embedded into the country’s R&D system as a strategic neural network.

Ultimately, a winning national system requires three things at once: innovation agencies capable of crossing the Valley of Death, AI-driven decision-making structures and legal guarantees of autonomy beyond rigid bureaucratic control. Korea once excelled as a fast follower running efficiently along paths pioneered by others. But in a world with no clear guide ahead, survival depends on the ability to create new paths independently. The future of Korea’s technological leadership will depend less on individual technologies than on whether the country can design a system capable of sustaining innovation itself.

This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom.