People often say, “We have the technology, but there’s no market.” That is wrong. The market is not absent; it is unseen — because the problem has not been defined.
South Korea has learned to run fast on roads others built — in semiconductors, displays and batteries. It has less often created the road itself. Artificial intelligence is a roadless technology: It can detect tiny factory defects, hidden bottlenecks in logistics routes, or whether an older adult living alone took medication — areas that were hard to see and therefore rarely treated as “problems.” When AI makes them measurable, they become problems, and once they are problems, markets open.
I recently met the CEO of an AI company in manufacturing. The firm proved its technology through a government project and had a model ready for the shop floor. The obstacle came next: finding factories that needed it and explaining its value in the language of the workplace — tasks it could not do alone. This is not unique. Many AI firms stall at the market’s doorstep. That is a bottleneck for South Korea’s AI industry.
Policy now needs to shift from funding technology to defining problems. Economist Mariana Mazzucato called this “mission-oriented innovation policy.” “We will foster AI” is different from “We will cut emergency response time for older adults living alone by half.” The first attaches money to technology; the second attaches technology to a problem. The logic is similar to the U.S. Defense Advanced Research Projects Agency using a single autonomous-driving race to spur an industry, or the European Union aligning technology and capital under the Green Deal. The clearer the problem, the easier it is for technology to find its destination, for companies to forecast returns, and for money to move.
Two steps are needed.
First, set numeric targets and concentrate support on companies that can deliver. Policy should start with challenges drawn from the field — defects in semiconductor back-end processes, farmland yields, emergency response for older adults. Demand should go to the few with the capacity to solve the problem. Spread support evenly and no one grows big.
Second, once companies are selected, support must be sustained. Until now, assistance has been fragmented by project and by year, failing to back growth. Computing resources, data, real-world testing, financing and overseas expansion should be linked, with support scaled up when results appear.
Some countries are already producing results this way. France, through French Tech Next40/120, selects 120 scale-up companies each year for full national support and links that effort to deep-tech development through French Tech 2030. Singapore’s Enterprise SG Scale-Up has 집중 supported more than 80 companies and generated S$2.5 billion in new revenue in three years. Israel, under a national AI program, has built a dedicated supercomputer, while its Innovation Authority directly supports companies with research foundations and funding. The methods differ, but the common point is clear: pick companies that can win and back them to the end.
South Korea faces the same task. To carry this approach into the field, it needs a small- and medium-sized enterprise-focused institution with both the ability to select firms and the policy tools to support them at each stage.
In the end, how precisely a problem is defined determines the size of the market. Technology flows in proportion to that size, and industry grows with it.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.