Investors' perspectives on the artificial intelligence (AI) market are changing rapidly. Just 2-3 years ago, AI often referred to generative AI services like ChatGPT or software innovation. However, recently, Nvidia CEO Jensen Huang has begun describing AI not merely as a simple technology, but as a massive industrial ecosystem and an industrial revolution.

The core concept he introduced is the 'AI Five-Layer Cake.' Through Nvidia's official blog and a World Economic Forum (WEF) dialogue in Davos, Jensen Huang explained the AI industry using a structure composed of five layers. He emphasized that "AI is not just one model, but a full stack," highlighting that AI is a gigantic industrial structure connecting power, factories, semiconductors, data centers, communication networks, and even robots.

In particular, the market's attention is shifting not to the top layer of the cake, but to the 'very bottom layer.' The following is a look at each layer of the AI five-layer cake defined by Jensen Huang and its investment implications.

Stage 1: Energy (The Bottom Layer) The foundation of the five-layer cake is energy. While many investors look to GPUs as the core of AI, Jensen Huang points to electricity as the most crucial element. AI data centers consume immense amounts of power, incomparable to traditional internet servers. For large language models (LLMs) to process billions of requests in real-time, hundreds of thousands of GPUs must operate simultaneously, requiring a stable power supply. Jensen Huang pointed out that the next bottleneck in the AI industry could be electricity, not semiconductors. In fact, in global financial markets, nuclear power companies, transformer manufacturers, and power transmission and distribution infrastructure companies are gaining attention as new AI beneficiary stocks. The influx of capital into nuclear power-related stocks and power infrastructure stocks in the domestic stock market, including Doosan Enerbility, is also driven by this trend.

Stage 2: Chips (Semiconductors) The heart of the current AI industry is the GPU. Nvidia is widely regarded as the company that has practically dominated the global AI GPU market. However, the chips Jensen Huang refers to are not limited to GPUs alone. They encompass the entire semiconductor ecosystem necessary for AI computing, including HBM (High Bandwidth Memory), advanced packaging, network chips, and AI accelerators. For this reason, Samsung Electronics and SK Hynix are considered key beneficiary companies in the global AI investment cycle, and the importance of HBM is growing to the point where it is called the 'oil' of the AI era.

Stage 3: Infrastructure (Data Centers and Networks) Moving to the infrastructure stage, the market size grows even larger. As AI advances, the amount of data movement increases exponentially, expanding the importance of data centers, networks, and optical communication infrastructure needed to process it. The AI industry is not completed by semiconductors alone; it leads to a competition to build massive infrastructure that quickly moves and stores data. This is also why global Big Tech companies are currently engaging in a massive AI data center investment race worth hundreds of trillions of won.

Stages 4 & 5: Models and Applications The upper layers consist of large language models (Models), such as ChatGPT, which the general public commonly thinks of as AI, and the services (Applications) utilizing them.

Investment Implications: Why Look at the Bottom Layer? Interestingly, Jensen Huang views the lower layers—energy, chips, and infrastructure—as more important than the models themselves. This is because no matter how excellent a model is, it is impossible to maintain competitiveness without the power, semiconductors, and data centers to drive it.

Ultimately, the market has begun to view AI not as a mere technological innovation, but as a new industrial revolution that transforms electricity, steel, factories, semiconductors, communication networks, and the robotics industry as a whole. The reason investors have started closely watching the bottom layers, such as energy and infrastructure, is the spreading consensus that the future value of the AI industry may be determined first by the foundational infrastructure of the lower layers rather than the services of the upper layers.

[※ This article was written for the purpose of providing investment information, and the investment decision and responsibility lie with the investor. AI was utilized in the process of writing this article.]