HBM Shortage Until 2028... Continued AI Investment

According to this, LG Energy Solution is expected to have 120GW, Samsung SDI 60GW, and SK On 15GW of production capacity.

Kim Jang-yeol stated, "LG Energy Solution's capacity by 2028 is 120GW. SDI is half of that at 60GW. SK On is 15GW."

Despite numerous positive reports on secondary battery-related stocks being published after SK On's large-scale order, their stock prices turned downward immediately after the reports were released. Kim Jang-yeol analyzed this phenomenon as a short-term correction due to profit-taking following a sharp rise.

He explained the market's reaction, stating, "The stock price was reflected yesterday. The moment the reports came out, everything dropped."

He explained that for POSCO Future M, the first selling point was when it neared 250,000 won, and for L&F, a correction was inevitable when the target price deviation came within 10%.

Kim Jang-yeol stated, "Today, the moment it hit 206,000 won, there was a first check. It's the same for L&F. The target price deviation came within 10%. In that case, a correction is inevitable."

He stated that it is difficult to immediately agree to apply a multiple of 10x, as BH heavily relies on the foldable phone momentum of Samsung and Apple.

He also pointed out that claims of the current stock price being undervalued should be viewed based on next year's figures. Furthermore, he added that BH's ability to reflect new momentum, such as in the robot component business, will be key to its future growth.

Kim Jang-yeol analyzed that if Samsung SDS's robot orchestration service successfully settles in, it could generate 30 to 50 billion won in sales per site within Samsung Electronics alone. This means significant revenue could be generated if about 80,000 robots operate across approximately 35 sites.

He further explained that Samsung SDS is transitioning from a traditional IT service provider to a cloud business structure centered on data center leasing, forecasting that this change will be a new growth driver.

Kim Jang-yeol predicted that the HBM (High Bandwidth Memory) supply shortage would continue until next year (2025). However, he emphasized that 2026 would be a crucial turning point to determine whether the supply shortage would continue long-term.

He added that how smoothly foundry companies can supply base dies will be a key variable for HBM supply.

He explained that if the token demand growth rate remains below 10 times, calculating the return on invested capital (ROIC) would be meaningless.

Conversely, he warned that if the overall industry's token growth rate surges to 100-800%, prices could fall, reducing profits, which would be a signal to be highly cautious about.

Kim Jang-yeol pointed out that there is also a critical point for data center investment. He warned that if borrowing rates, currently in the 6% range, increase by more than 2 percentage points, it would significantly impact investment.

He also emphasized the need to distinguish between state-level and federal-level regulations regarding data centers. He analyzed that federal-level regulations would face political constraints before becoming a reality.

However, he emphasized that this does not mean an oversupply, but rather that the supply and demand situation could improve compared to now.

He added that a new investment cycle (round) is expected to begin between 2029 and 2030, and this period will be a crucial variable in determining the direction of the HBM market.

He believed that if agents perform various tasks such as coding, scheduling, and research, and even merge with each other, token demand will explode.

He explained that just as Google experienced a 15-fold increase in tokens in the past, the increase in token demand due to the proliferation of AI agents could normalize. He analyzed that the market's current skepticism about AI monetization is the main cause of stock price volatility.

Kim Jang-yeol estimated that private companies could reduce labor costs by about 30% by adopting AI. He analyzed that companies would use a significant portion of these saved labor costs for AI reinvestment, which would, in turn, lead to increased demand for cloud services.

This virtuous cycle of AI-based labor cost reduction and reinvestment is expected to drive the continuous growth of the semiconductor industry. He explained, "You can cut personnel. Private companies say labor costs account for 30%, and a considerable portion of those labor costs, about 30%, will be reinvested as companies cut staff to operate their businesses, and that consensus is growing."

He emphasized that if companies expecting 50-70% growth next year (2025) are valued at a price-to-earnings ratio (PER) of 15x or less, no AI-related company worldwide would be recognized for its investment value.

This analysis suggests that Nvidia's current stock price reflects the future growth potential of the AI industry, and as long as future earnings growth continues, the possibility of a stock price crash is low.

Kim Jang-yeol explained that Apple uses more than its free cash flow for shareholder returns, allocating most of it to share buybacks. It uses a strategy of retiring 85% of the repurchased shares to increase per-share value.

He analyzed that these repeated share buybacks and retirements consistently maintain book value and build trust in the company, acting as a factor that defends against stock price declines.

Kim Jang-yeol mentioned that the market's supply-demand imbalance is likely to resolve around October 13. Samsung Electronics and SK Hynix are scheduled to announce their Q3 earnings in mid-October, and these earnings announcements are expected to act as new momentum.

He analyzed that institutional and foreign selling might continue until mid-October, but if Q3 earnings meet market expectations, the supply-demand situation will turn positive.

LIG Nex1's current order backlog is approximately 25 trillion won.

He analyzed that this increase in order backlog has the potential to boost operating profit by 40-50%. He added that the profit margin for overseas orders, at mid-20%, is significantly higher than domestic orders, which will greatly contribute to profit improvement.