Nvidia is Risky, Where Does the Money Go in the AI Era?

Nvidia recently captured market attention by projecting a 70% annual revenue growth rate in its latest earnings announcement. This figure significantly surpasses the market consensus of 44%. The management even stated that this was a conservative estimate, showing high confidence.

CEO Jang Woo-jin pointed out that despite Nvidia presenting an enormous level of guidance, the stock's increase was limited, unlike past earnings surprises where it surged by over 15%. He analyzed that the investor sentiment is not fully reflected, as the stock only rose about 4% in the days following the earnings announcement.

Questions about the profitability model of AI companies are growing. AI companies like OpenAI have a structure where even if revenue increases, the costs for using cloud services are very high, preventing it from translating into profit. This is because the semiconductor and infrastructure costs essential for AI model development and operation are substantial.

Currently, the AI market is experiencing intensified competition, leading to a 'chicken game' where service prices are being cut. In this situation, the high variable cost structure of AI businesses makes it difficult to generate profit despite revenue growth.

Consequently, concerns are being raised about the source of funding for AI companies to cover cloud costs and achieve sustained growth.

Unlike semiconductor companies, AI service companies have a high variable cost structure. For example, AI services like ChatGPT have the characteristic that as usage increases, cloud usage costs also increase. In such a high variable cost structure, if service prices are lowered, it is difficult to make a profit even with increased sales, likely leading to continuous losses.

Building AI infrastructure requires enormous amounts of power and water, and there is a risk of delays in related permitting processes. Currently, memory semiconductor prices are rising due to expectations of AI infrastructure expansion, but if actual infrastructure construction is delayed, pre-purchased semiconductor inventory could accumulate, leading to price drops. This increase in inventory could pose a significant risk to the semiconductor market around 2028.

While past big tech companies succeeded by securing dominant market positions, OpenAI is criticized for not yet establishing a dominant position despite high investment costs. OpenAI's LLM (large language model) has attracted massive investment, but its profitability remains questionable. Moreover, instances of competitor Anthropic's model performance surpassing OpenAI have emerged, diversifying market expectations.

Jang Woo-jin analyzed that Google's Gemini AI is deeply integrated into Google's existing services, making its disappearance unlikely. Gemini aims to differentiate itself by embedding AI features within Google's search engine, enhancing the user's search experience. This suggests that success will depend not merely on the performance of the AI model itself, but on how effectively AI is integrated into a company's business ecosystem.

This strategy leverages Google's existing vast user base and data to expand AI services. Ultimately, the AI market is transitioning from a competition of individual model superiority to a battle of business ecosystems that combine with broader services to provide value to users.

Therefore, Google is expected to focus on strengthening the competitiveness of its existing services through Gemini, preventing user churn, and creating new revenue models.

Jang Woo-jin pointed out the current lack of a 'moat' in the AI market. A moat refers to competitive advantages that prevent competitors from entering and maintain a dominant position. He explained that unique user experiences that prevent churn, like Google or Naver's Knowledge iN in the early days of the internet, have not yet appeared in the AI sector.

Currently, AI services offer similar functionalities and lack strong differentiators that would make users dependent on a specific service. This means users can easily switch to other AI services at any time, ultimately making it difficult for a winner-take-all structure to form.

However, Jang added that if a dominant leader emerges offering an overwhelming user experience, the current market structure could rapidly reorganize.

Excessive capital concentration in the AI industry could lead to a credit crunch in other sectors. Under high interest rates, if investment is focused solely on AI, there's a warning of a K-shaped polarization where marginal firms and households face a heightened risk of collapse. Jang Woo-jin diagnosed that money concentrating in specific industries amidst high interest rates would exacerbate risks for the global economy.

South Korea, with its export-oriented economic structure, may experience a relatively limited impact on exports from domestic interest rate hikes. However, domestic industries, especially retail sectors like department stores and supermarkets, will be negatively affected. Ultimately, Jang Woo-jin predicted that a prolonged high-interest-rate environment could lead to a general decline in corporate valuations.

South Korea exhibits characteristics that make it relatively less sensitive to global inflation shocks. This is due to the Korean Won maintaining strength compared to other currencies and the country's possession of domestic refinery facilities. These domestic refineries contribute to stabilizing energy prices and create a stable supply and demand environment, even amid external shocks like global diesel shortages.

Recently, global diesel crack spreads have surged to levels similar to the $120 oil price during the 2022 Russia-Ukraine war. This is due to disruptions in global diesel supply caused by ongoing attacks on Russian refinery facilities.

Russia accounts for a significant portion of global diesel exports, and refinery attacks are causing cascading supply shortages not only in Europe but also in Asian markets. This has led to gasoline shortages in neighboring countries like Mongolia, raising concerns about energy security.

Especially as winter heating demand increases, diesel demand will rise further, making the current supply instability a factor that could exacerbate global inflationary pressures.

This surge in diesel prices could lead to increased logistics costs, resulting in higher overall commodity prices.

Jang Woo-jin predicted a high possibility of a market correction in September. The U.S. Treasury plans to focus on replenishing its Treasury General Account (TGA) balance until September and begin releasing funds into the market in October. Furthermore, September statistically tends to be a weak month for the market, and major rebalancing by asset managers after summer holidays could also increase volatility. The expectation of lower third-quarter earnings also contributes to the likelihood of a September correction.

In the event of a September market correction, Jang Woo-jin recommended index-focused investing rather than individual stock investments. He explained that it is difficult to respond by selecting individual stocks, and buying KODEX Semiconductor ETF or broad index ETFs could be a safe strategy.

He advised buying leading stocks when they fall, or investing in the entire index for peace of mind, to cope with market volatility.

This judgment is based on the idea that in times of high uncertainty, it is more advantageous to ride the overall market trend rather than focusing on individual corporate risks.

Jang Woo-jin recommended allocating about 30% of a portfolio to long-term bonds and gold. Long-term bonds and gold act as insurance assets, providing a hedge effect during recession fears. He explained that while gold prices might fall if recession concerns grow, falling long-term bond yields can help protect the overall portfolio from losses.

The biggest variable determining the market's direction in September is interest rate volatility. If interest rates surge as they are now, the market could be significantly shaken again, requiring caution. Additionally, CPI (Consumer Price Index) and PPI (Producer Price Index) data, to be released in early September, will also have a significant impact on the market.

The non-farm payrolls report, in particular, is considered a crucial economic indicator. Jang Woo-jin emphasized that after the temporary employment increase effects from past events like the World Cup disappear, the non-farm payrolls and unemployment rate figures will provide insight into the actual health of the labor market.

These indicators are key grounds for the Federal Reserve's (Fed) monetary policy decisions, so investors should pay close attention to changes in these figures.