Cautious Japanese Companies Begin Expanding Factories

They warn that the current stock price increase may be an unsustainable bubble, and such concerns are repeatedly raised in financial broadcasts.

These companies raised their guidance in Q2 earnings announcements and signaled large-scale capital expenditures, showing an unusually aggressive attitude.

Unlike Korea, the Japanese market has a diversified structure where market influence is not concentrated in IT companies. Japanese banks, especially the three major banks (Mitsubishi UFJ, Mizuho, and Sumitomo Mitsui), have well-diversified global assets, earning more than half of their total revenue from overseas.

This acts as a structure where bank stocks benefit during periods of rising interest rates, showing that profitability can fluctuate depending on global economic conditions.

Indeed, even during a sharp decline in IT stocks, value stocks such as banks, trading companies, and real estate maintained solid performance, playing a role in enhancing market stability.

Japanese companies are not only raising their earnings guidance compared to three months ago but also announcing aggressive expansion plans across the entire AI supply chain. This is an unusual move for Japanese companies, which have historically shown a conservative investment tendency, and is interpreted as reflecting their confidence in actual demand growth.

Advantest, the world's No. 1 company in semiconductor post-processing test equipment, recently reached a new high and entered Japan's top 10 in market capitalization. The company significantly raised its annual forecast for this year during its quarterly earnings announcement in late July and showed an independent upward trend even amid declines in other semiconductor companies like Kioxia.

Several Japanese component companies, previously considered conservative, are attracting attention by announcing large-scale expansion plans related to AI. Mitsui Kinzoku is pushing for a major expansion of copper foil for semiconductors despite earnings below market consensus.

Furukawa Electric provided a positive outlook, including a 45% upward revision of its net profit forecast.

Ibiden also projected increased profits despite sluggish smartphone and automotive markets, suggesting that AI-related demand is driving overall performance.

Demand for data center investment is strong, but shortages are occurring at physical production stages for electricity, semiconductors, components, and equipment.

This is due to structural limitations that make it difficult to quickly double or triple the production capacity of goods.

Therefore, it is analyzed that the biggest constraint in the current AI industry is the supply of physical infrastructure, rather than software development.

The correlation between Nvidia's strategy of strengthening NVLink and its memory utilization strategy should also be noted. Analysis suggests that if optical communication capabilities improve during computing, memory dependence could decrease.

This is an important indicator suggesting that actual demand growth linked to the AI industry is leading to data center infrastructure investment.

Japanese manufacturing also showed a strong recovery, with new orders in August reaching their highest level since January 2018.

As it becomes difficult to secure the necessary workforce for on-site construction, the adoption of modular data centers is increasing.

The modular approach involves assembling pre-fabricated modules at the site, which can reduce on-site labor and shorten construction periods.

Even if it costs more, it helps solve labor shortages and improve overall construction efficiency, making it an alternative to overcome the limitations of on-site construction.

Indicators for U.S. regional banks are showing positive signs of real economic recovery. In particular, the U.S. Regional Bank ETF (KR) has shown a robust performance, rising 16% year-to-date.

This indicates that banks are maintaining profitability through increased lending and fee income, despite rising interest rates.

This phenomenon implies that the trickle-down effect of economic revitalization is spreading not only to large corporations but also to local economies.

In Korea, AI-related investments are concentrated around a few large corporations like Samsung Electronics and SK Hynix, potentially leading to technological bottlenecks. In contrast, in the U.S., infrastructure investments such as data center construction are spreading nationwide, allowing for a broader trickle-down effect.

While Japan does not have large corporations like Samsung Electronics or Hynix, it possesses numerous small and medium-sized materials, components, and equipment (MCE) companies that are strong in their respective sectors. These companies are making diversified expansion investments across the entire AI supply chain, a characteristic that leads to broader economic trends.

This diversified investment structure in Japan, instead of relying on one or two companies, drives growth simultaneously across various sectors, suggesting that economic effects can spread to a wider range of businesses and regions.

In the Japanese investment market, there is still a heterogeneous reaction to the rapid rise of IT stocks. Experts and media in Japan perceive the surge in AI and IT stocks as a bubble and continue to warn about overheating.

Some individual investors even dismiss the surge in IT stocks as similar to Korea's thematic stock investments, expressing difficulty in understanding it.

Despite these warnings from institutions and the media, the overall upward trend of IT stocks continues, showing a divergence between perceptions within the Japanese investment market and actual market trends.

Analysis suggests that the market is currently underestimating the demand for AI data centers. The AI industry is not a simple short-term trend but is in a long-term flow of change, to the extent that it should be approached as essential basic infrastructure, like railways, electricity, and telecommunications.

Experts believe that the AI industry has just entered the early stage of a 10-year growth cycle.

This emphasizes that it is a massive change that should be viewed from a long-term perspective of 100 years, beyond short-term demand over three years.

Typically, companies tend to present such forecasts cautiously.

However, currently, a large number of companies, not just one or two, are collectively issuing aggressive forecasts.

This suggests that it is not simply a short-term, one-year investment, but a move towards long-term physical infrastructure expansion. The collective optimism of companies is interpreted as a demonstration of confidence in the AI industry's growth potential.