● Personal thoughts on the semiconductor sector

[Summary of past semiconductor sector stock price flow]

The chart is the mainstay of the Philadelphia Semiconductor Index
When the index’s weekly chart breaks strongly above the 20-week mark will be the start of a rally in the semiconductor sector. If it fails to break through, it will be similar to what it was in February 2022
February 2022
In the process of switching to contact in the context of the COVID-19 pandemic, IT demand has declined and big tech’s margins have begun to fall. Order cuts from set companies and inventory surges from memory companies appear. Rapid inflation begins, leading to a sharp rate hike cycle. Starting with a 25bp hike in March, 425bp hike in May, 75bp in June, 75bp in July, 75bp in September, 75bp in November, and 50bp hike in December, the benchmark interest rate will reach 4.25-4.5% in 2022. It reaffirms that the semiconductor sector will be sluggish in a bad macro situation. The U.S. CPI peaked at 9.1% in June
October 2022
In 3Q results, memory inventories recorded a peak out, and the preemptive stock price bottom was indicated by the announcement of production cuts during an earnings call of memory companies. The peak out of inventory has emerged along with the possibility of limiting supply due to the increase in the difficulty of process miniaturization of DRAM. From this time on, the market’s interest in post-process chiplet technology began to increase. Warren Buffet bought TSMC during this period
January 2023
The effects of memory production cuts have been confirmed since the first quarter, and uncertain mood continues before confirmation of supply and inventory declines. Eventually, demand growth was needed, but OpenAI’s ChatGPT emerged, and Nvidia’s GPGPU, HBM, and TSMC’s CoWoS packaging began to emerge
October 2023
Nvidia was also the one that saved the semiconductor sector, which had been undergoing adjustment due to the macro situation since August. The peculiar thing is that Qualcomm’s stock price rise was good, but it began to expect the spread of the AI cycle with on-device AI. The period when the AI semiconductor value chain’s interest spread
April 2024
HBM’s trickle-down effect has raised expectations for the legacy semiconductor cycle. Samsung Electronics recorded KRW 6.6 trillion in its first-quarter earnings, exceeding the consensus of KRW 5.3 trillion, and KRW 10.4 trillion in the second quarter, exceeding the consensus of KRW 8.3 trillion, continuing its expectations for a recovery in demand for legacy semiconductors. Interest in the small-scale department of the entire process increases during this period
As of September 2024
From the end of June, the global research house began to undercut the tech sector due to the macro situation caused by the economic slowdown. As the time for a rate cut approaches, concerns about a recession grow. Eventually, there was a sharp decline in the tech sector following the liquidation of Encarry trading. Recently, concerns such as oversupply of HBM, increased memory inventory and prices, falling margins, and delayed recovery of demand for legacy semiconductors have been added

[Thinking about semiconductor stock price]

Recently, a rather unusual report on the memory industry came out, such as a double downgrade of SK Hynix’s investment rating. I heard that there was also malicious intent in the inside story, but aside from that, I will give my personal opinion on the main thesis of the report.

First, there is an opinion on the oversupply of HBM. The opinion is that overinvestment centered on big tech is not sustainable, and that the expansion of supply from competitors such as Samsung Electronics and Micron could change from SK Hynix-led to oversupply, which could lead to a decline in margins. In the past, cost reduction was a major issue in the semiconductor industry because the price decline was faster than the performance increase. This is the case for general-purpose semiconductors and does not take into account the specificity of HBM. The minimum workstation or supercomputing uses HBM. HBM is packaged on top of processes and supplied to Nvidia and others. Even if only one HBM fails, the yield is a very important issue because the entire H100 has to be thrown away, and to be precise, the optimal yield for my product is important. This is also the reason why Samsung Electronics has not been able to enter the market properly yet. While big tech companies such as M7 are lining up to develop HBMs customized for their accelerators, the level of competitors is not rising so fast to discuss oversupply. If we go to HBM4 after next year, customized HBM will become more important, and the more customized it is, the more the price negotiations change to be supplier-oriented.

There are concerns that Samsung Electronics’ performance in the third quarter will be bad. This seems to be a situation where it is trying to rob costs in the third quarter, so we judge it as true. The consensus for operating profit exceeded KRW 14 trillion, but recently it seems to be falling below KRW 10 trillion. The decline in the consensus of KRW 4 trillion or more is a matter of reflecting costs due to yield problems and not paying employee incentives, but setting provisions should be set more than expected. I understand that the atmosphere is unusually high in the third quarter due to political implications, but aside from the details, if we announce results of KRW 10 trillion or more, it is highly likely that the market will not reflect it any worse. The important thing is the performance from the fourth quarter. If we have certainly robbed a lot in the third quarter, we just need to instill expectations that the fourth quarter will be able to recover to its normal performance. Internally, Samsung Electronics is still seeing an operating profit of KRW 50 trillion in 25 years based on semiconductor performance alone. If we give an operating profit of KRW 10 trillion in the early part of the third quarter and KRW 13 trillion in the fourth quarter, the PBR based on ROE of 10% in 24 years will be approximately 1.15 times higher, and the PBR will fall to KRW 1 times higher considering the expected performance in 25 years. Since the PBR has never been below 1x except for the 2008 financial crisis, the bottom value basis seems sufficient. The question will be at what point can we expect a stock price rebound.

Starting in the second quarter of this year, Samsung Electronics has attempted reorganization, personnel measures for executives and employees, and revamping the atmosphere. I think the results of the change should be at least in the first and second quarters of next year, and I expect a very different response from stock investors when summer comes next year.

tslaaftermarket