AI Investment Boom Continues…Opportunities in Small and Mid-Caps

During the market decline in June-July of last year, concerns about AI investment halts and prolonged high interest rates spread throughout the market.

Putting together the facts confirmed so far, AI investment is still continuing actively. The current index is going through a process of absorbing past overhead supply within a box range around the 7,100 level.

Ahead of Chuseok, large-scale selling by individual investors occurred, and trading volume dropped significantly compared with April-June.

With no strong buyers in the market, treasury share purchases by other corporations stand out and are supporting the market.

Market supply-demand, lacking the strength to lift large caps, is showing movement toward mid-caps and small caps rather than ultra-large caps.

The key driver currently supporting the 7,000 level depends on the direction of next year's annual operating profit for all KOSPI companies.

Based on the operating profit trend so far, the 7,000 level is sufficiently explainable, and considering export data hitting record highs, current stock prices cannot be considered expensive.

VKOSPI, the indicator representing KOSPI volatility, exceeded 90 in June and has now fallen to the mid-40s.

This is still high compared with last year's average of around 20, but opinions are emerging among institutional investors that new buying restrictions will be lifted only when the volatility indicator stabilizes.

Stock prices have fluctuated, but real indicators such as actual orders and order backlog remain solid.

In Q2 earnings and Q3 earnings previews, no real changes related to AI investment have yet appeared.

A benchmark interest rate of 5% means it takes 20 years to recover the principal, similar to a stock's P/E of 20 times.

When deposit rates were low, even overvalued companies were preferred as alternatives, but in a high-rate environment, controversy over high-PER companies is inevitable.

Even for companies generating the same earnings, it is reasonable to lower the bar for corporate valuation (multiples) in line with changes in the market environment.

During the COVID-19 pandemic, liquidity rallies made it possible to invest by looking only at whether prices would rise.

Since the U.S. rate hikes in 2022, companies that actually generate profits and those that do not have become clearly distinguished, and while there are companies that make money even in a high-rate environment, the market's criteria for identifying them have become much stricter.

High interest rates and inflation raise companies' material and procurement costs, eroding operating profits.

Companies that can pass cost increases on to product prices can achieve both sales and profit growth at the same time, but companies that cannot pass on prices and also see declining sales volume enter a profit-decline cycle.

In an environment where costs are rising worldwide, countries where AI investment takes place see investment, employment, and consumption occur in a chain, showing a national economy that can bear the costs.

On the other hand, Europe and others lacking AI investment do not have the capacity to bear the costs and are placed in a relatively disadvantageous situation.

Korea has the product lineup needed for the AI cycle, so it can seize a big opportunity within this trend.

In early July last year, news of Meta's data center leasing was interpreted as the starting point of AI semiconductor leverage liquidation, raising concerns in the market about reduced AI investment.

However, Mark Zuckerberg explained it as efficiency through spec upgrades, not a lack of demand. As new demand such as AI agent services is being confirmed recently, short-term negative opinions can be viewed as mere noise.

The market was temporarily shocked by arguments from Anthropic and others about slowing the pace of AI model development.

However, major companies such as Meta and OpenAI continue to release new models, and OpenAI's GPT-4.0 showed better performance metrics than previous models. Contrary to concerns in some parts of the market about reduced investment, the actual pace of development is not slowing.

Doubts were raised about the performance superiority of Chinese AI, and Anthropic announced that Chinese AI models did not learn from its models but 'copied' them.

During the Q&A process, even national confidential information was leaked, showing the limits of evaluating investment value based on simple performance metrics alone.

In the current AI field, large-scale investment is taking place over a historically unprecedented short period, and national-level investment is occurring as part of the U.S.-China hegemony competition.

Big tech companies' AI service purposes still remain at the level of simple search, but like the case of exploding data traffic in the mobile era, the technological revolution is expected to lead to even greater changes.

The emergence of such services is a starting point that makes it possible to predict changes in users' lives, suggesting that the current technological stage is not the end of the revolution but a process of a productivity revolution.

Companies already verified in the market through the AI cycle over the past 2-3 years, delivering products and generating revenue, still maintain leadership in the AI cycle.

Rather than newly emerging rising star stocks, attention should be paid to the earnings of existing suppliers already receiving orders from big tech companies, and in the AI cycle, the winners have already been decided.

In the semiconductor sector, it is difficult to expect new upstart companies or sudden technological innovation. Among companies that have already entered vendor lists, it is important to endure the tedium of the revenue recognition period.

Companies expanding production capacity (CAPA) demonstrate management's confidence in winning orders, and such companies deserve attention.

Through the first half of last year, profit growth driven by rising semiconductor prices (P) for HBM, DRAM, NAND, etc. led the market.

As we move toward the second half of 2026, the market is expected to enter a phase where volume (Q) increases, and accordingly interest is shifting from finished semiconductor companies to the materials, components, and equipment sector.

This year, KOSPI returns recorded over 60%, but KOSDAQ still remains in negative territory for the year.

The rallies in August-September largely resolved supply-demand distortions caused by single-stock leveraged ETFs, and this is a phase in which the unusually wide stock price gap between large caps and small/mid-caps is being closed.