[Edaily Reporter Kim Hyun-ah ] In the 1970s, the engine driving South Korea’s economy was the heavy and chemical industries. Massive factories producing steel, building ships, and manufacturing automobiles led the nation’s growth. At that time, the country’s competitiveness depended on how many factories it could build and how many products it could manufacture and export to global markets.

Now, some 50 years later, the world is once again facing a massive industrial transformation. This time, it’s neither steel nor automobiles. “AI Factories”—factories that produce intelligence—are emerging as a new source of national competitiveness.

This is also the message that SK Group Chairman Choi Tae-won and former Senior Presidential Secretary for AI Future Planning Ha Jung-woo both conveyed in connection with the government’s recent announcement of the “Three Mega Projects for Korea’s Great Leap Forward.” They emphasized that AI data centers (AIDCs) are no longer simply server rooms for storing data, but factories that produce intelligence.

From the Era of Manufacturing Goods to the Era of Creating Intelligent Services

During the era of heavy and chemical industries, factory structures were simple. Raw materials such as iron ore and oil were fed into blast furnaces and production facilities to create products like automobiles and ships. These products were then exported overseas, driving economic growth.

The principle behind AI factories is not much different. The raw material is data, and the energy is electricity. The core production equipment that processes these is AI semiconductors such as GPUs and HBM. AI models perform training and inference on this equipment to generate AI services such as translation, design, coding, new drug candidate discovery, and robot control.

In this process, the outputs generated by AI are processed in units called “tokens,” which are traded and billed based on usage. This is why global AI companies today sell AI services based on token usage. Just as factories in the past produced automobiles, AI factories are essentially factories that produce and supply intelligent services in units of tokens.

The AI Factory is the “factory” of all factories in the world

. While heavy and chemical industries fostered the growth of specific sectors such as steel, shipbuilding, and automobiles, the influence of the AI Factory is far broader. AI is a universal infrastructure that boosts productivity across all industries, including manufacturing, finance, biotechnology, defense, healthcare, and content. Referring to this, former KAIST President Lee Kwang-hyung described AI as “akin to the discovery of fire.”

In shipyards, it controls welding robots; in semiconductor factories, it optimizes designs; and in pharmaceutical companies, it accelerates the search for new drug candidates. In finance, it performs risk analysis, and in hospitals, it supports medical staff in making diagnoses. Just as factories of the past produced goods, the AI Factory can be described as the “factory of factories” that boosts the productivity of every factory in the world.

Having Lots of GPUs Doesn’t Make a Country an AI Powerhouse

So, can a country become an AI powerhouse simply by securing a large number of GPUs? Of course, the government’s decision to secure 260,000 GPUs by 2030 is significant in terms of securing production infrastructure. However, former Senior Secretary Ha warns, “A data center filled only with GPUs could degenerate into an AI subcontracting hub that merely performs computations for overseas Big Tech companies.” He explains that this could create a structure where Korea provides only electricity and land, while overseas companies reap the actual added value.

The key point is what kind of AI services are produced. AI used for everyday chatbots is entirely different in the value it creates from AI that controls welding robots in shipyards or designs semiconductors—even if they use the same GPUs. Depending on what industrial data is used for training and which models are deployed, the economic value can vary by tens or even hundreds of times.

That is why AI yield is so important. The ability to produce more accurate and valuable AI services at a lower cost—while using the same power and the same GPUs—determines a country’s AI competitiveness.

Competitiveness arises when infrastructure and software work together

. At this point, the messages from Chairman Choi Tae-won and former Senior Secretary Ha converge.

Chairman Choi emphasizes the industrial infrastructure needed for AI factories. He argues that we must build large-scale power grids, construct AI data centers, and secure supply chains for AI semiconductors, including HBM. On the other hand, former Senior Secretary Ha emphasizes the production processes that will run within those factories. His strategy is to train AI using the industrial data accumulated by Korean manufacturing and combine proprietary AI models with AI operations (AI Ops) to create higher value at the same cost.

AI competitiveness is not determined by the size of the factory alone. Only when the hardware—the infrastructure—is combined with the software—AI models, industrial data, and AI Ops—can an AI factory become competitive in the global market.

An Era of Competing on “Intelligent Productivity” Rather Than Steel Output

In the 1970s, nations competed over steel

output

and automobile exports. That was a measure of national power.

What will determine national competitiveness in the future? It will not be how many GPUs a country possesses, but rather how much high-value-added AI services it can create using those GPUs and how much of that it can sell on the global market.

This is why Chairman Choi Tae-won refers to AI data centers as “AI factories” and “token factories,” and why former Senior Secretary Ha emphasizes AI yield and proprietary models over the sheer number of GPUs.

If heavy and chemical industries transformed South Korea into an industrialized nation 50 years ago, it is highly likely that AI factories will be the engine driving South Korea toward becoming an “intelligence exporter” over the next 50 years.

Now, the premier product to be exported to the global market goes beyond automobiles and semiconductors—it is “intelligence” itself, the force that drives industries. That is South Korea’s new growth engine in the AI era.