저자: 출연자미상 · 한국인
요약 보기
한줄결론: 발언자는 메모리 가격 사이클이 공급 조절에 의해 만들어진 것이며 인위적 급등이라고 보고, 삼성전자·SK하이닉스 주가의 적정가 회귀와 메모리 가격 하락 전환(빠르면 4분기, 늦어도 내년 1분기)을 예측한다. [시장 배경과 주가 흐름] 발언자는 자신이 본업은 주식이 아니라 산업 분석이라고 전제하면서도, 한국 증시가 2025년 4월 이후 급등했다고 설명한다. 코스피가 2,800에서 9,300까지 1년 만에 오른 것은 비정상적이라고 보며, 주가 상승의 원인을 산업 측면에서 찾아야 한다고 주장한다. [가격 메커니즘 주장] 발언자는 원유 사례를 들어, 수요가 늘어도 가격은 바로 오르지 않고 공급 이상이 있을 때만 급등하며 정상화되면 원래 수준으로 돌아온다고 주장한다. 메모리도 같다고 본다. 연간 가격 하락률이 20%를 넘으면 불안정, 20% 미만이면 나쁜 시장이라는 기준을 제시하며, 과거 30년간 메모리 가격 지표를 사이클로 그려보면 상승 사이클에서 가격이 오른 적이 두 번뿐이었다고 말한다. 2018년 슈퍼사이클 주장도 6개월 내에 가격이 급락하며 실패했다고 회고한다. [AI 수요 논쟁] 빅테크의 AI 수요 급증 주장에 대해 발언자는 객관적 데이터상 사실이 아니라고 반박한다. 데이터센터 수요는 전체 반도체 수요의 30~40% 수준이며 최근에도 50%를 넘지 않는다고 본다. 데이터센터 외 수요(PC, 스마트폰, 가전)가 나머지를 차지하며, 메모리 비트 성장률은 AI 도입 전 약 30%에서 현재 19%로 낮아졌다고 주장한다. 메모리 사이클은 수요가 아니라 공급에 의해 100% 결정된다고 단언한다. [수익성과 카르텔 의혹] 엔비디아 영업이익은 약 12배 늘었지만 메모리 기업은 생산량이 비슷한데 가격만 올랐다는 점을 지적한다. 메모리 영업이익률이 70~80%대에 달하는 점도 비정상으로 본다. 삼성전자와 SK하이닉스가 가격 유지를 위해 가동률을 70~80%로 억제하고 있다고 주장하며, 두 회사가 생산을 풀면 즉시 공급 과잉이 온다고 본다. 캘리포니아 연방법원에 제기된 집단소송(삼성, SK하이닉스, 마이크론 대상, 3배 배상 요구)을 언급하되, 담합 여부는 법원이 판단할 문제라고 선을 긋는다. [투자·치킨게임] 설비 투자는 현금 과세(미사용 현금 과세) 때문에 불가피하다고 본다. 메모리 3사 체제에서 중국 CXMT의 점유율 상승 등 과거 치킨게임 사례를 들어 업황 악화를 전망한다. 중국 반도체 수출이 2019년 이후 한국을 넘어섰고 중국 메모리 수요 비중이 약 60%라는 점을 지적하며, 중국 내재화로 공급이 더 늘어날 것이라고 주장한다. [전망] 발언자는 AI 수요 둔화가 내년쯤 시작되어 AI 버블 이슈가 본격화될 것으로 예상한다. 메모리 가격은 4분기 또는 늦어도 내년 1분기부터 하락 전환하고, 하락 시 급락할 것으로 본다. 적정 주가는 메모리 가격이 2025년 4월 이전 수준으로 돌아갈 때의 삼성전자 약 7만원, SK하이닉스 약 20만원 수준이라고 제시한다. 주가 바닥은 메모리 가격 바닥보다 먼저 올 수 있다고 덧붙인다. 정확한 시점은 가격 하락 속도를 봐야 알 수 있다고 유보한다. 이 발언은 본인이 전문가의 전망으로 제시하는 내용이며, 자막 자동인식 특성상 일부 용어와 수치는 부정확할 수 있다.
다듬은 글
I'm actually not someone who analyzes stocks, you know. It is about analyzing the industry. But well, stock prices are like a byproduct. It is essentially just a matter of naturally getting a sense of stock prices to some extent while predicting the industry. Yes. If you look at the timing when our country's stock market started to surge, yes.
It starts in April 2025. That's right . yes . That makes no sense. It surged, but before that, regarding the KOSPI, I’ve been in the stock market for 30 years. Since my student days. yes. yes. It’s really stagnant water, but on average, it took 8 to 10 years for the KOSPI to rise by 1,000 points . hmm.
hmm. But it didn't take a year to go from 2,800 to 9,300; it took exactly one year. I don't quite understand why anyone would consider that itself to be a normal condition. If so, we need to look closely at what happened in the meantime to cause the stock price to rise . Now, from an industrial perspective, memory prices started to skyrocket in April 2025.
yes. yes. Then, naturally, what goes up with it? profit. yes. But [snorting] as you know, regarding the price, I think you'll understand if you look at oil prices . Oil prices do not rise when demand increases, nor do they fall when demand decreases . hmm. If prices rise, they go up, or if supply skyrockets due to a war in the Middle East like now, and when things normalize, they return to the original $60 .
The price of the square is also the same. Prices only fluctuate rapidly when there is a temporary abnormality in supply, and compatible systems do not see price increases either. I have been doing this for decades and I understand that the difference between compatibility and instability is not whether prices go up or down, but that if the annual price decline rate is higher than 20%, it is instability.
hmm. If it drops by less than 20%, it is considered a bad market . So, the fact that the store opened is abnormal from the start. If you don't understand this in the memory market, you will understand if you look at the price indicators over the past 30 years . If you look closely at memory semiconductor price indicators and plot the cycles, prices have never risen when the cycle is rising.
The price went up exactly twice . You must know the situation in 2018 and now since we did a lot of broadcasts together . In the end, it turned out to be a bubble. Even back then, people said the supercycle would last five years, but less than six months after I spoke on TV, prices plummeted and the country suffered severe losses for two years.
It’s pretty much the same now. Prices do not rise like this under normal circumstances. The price only rises like this when abnormal situations occur in production . Even now, prices have risen in that way, but abnormally high prices will naturally return to their original levels once supply normalizes.
No, but as far as we know, it is true that demand driven by AI has recently skyrocketed. That is the company's claim. Ah, that is the company's claim. That is correct. No, based on objective data, that is not the case. So, for example, since numerous American companies known as Big Tech need to manufacture a massive number of data servers, they naturally require NBDI chips and the HBM that goes into them.
Isn't this just common sense? So, actually, what needs to be distinguished now is that the demand for data centers has increased significantly. yes. But data centers aren't the only source of memory demand . It's not just data centers. I do n't have data on the exact proportion of global semiconductor memory, but I have been observing the semiconductor market for several years.
yes. Now, if we consider the total semiconductor demand to be 100 days, data centers do not exceed 30 to 40 percent . usually. yes. Originally. But it recently surpassed that. It probably didn't go over . I don't think it will go over 50. It may have exceeded memory alone, but if we consider total demand, the remaining 50 to 60% or more is other financial sectors not related to data centers .
The demand for desktops, laptops, smartphones, home appliances, and the like is currently around 10% to 20% for each. Well, I guess that could be possible . Okay, so the demand for the product with a 40% share increased by 30%. yes. The proportion decreased by 15% from 60% . yes. If you calculate it, it comes out to about 10%, and that is the fact.
Actually, if Bit Growth continues to make announcements, I monitor them. AIBs started appearing in earnest around 2024, right? yes . Then, when looking at whether there was a change in the growth rate of overall memory demand compared to before, especially regarding DRAM and HBM since it is a DM, it actually decreased.
hmm. So, nothing affected the demand side . AI data centers are a fact, but what's more important is that memory cycles are our current concern. The real concern isn't so much how much demand there actually is, but rather that for memory companies to ultimately make money, the cycle needs to remain in a boom cycle.
Yes. Then, the premise that the memory cycle is determined by demand is absolutely necessary. hmm. That's right . yes. But that premise itself isn't a fact. Demand is that high, and the memory cycle has absolutely nothing to do with it . No matter how many times I plotted the data over the past 25 years, demand has always trended upward.
Severe failure. No matter how much of a deficit there is, demand continues to increase. Demand is increasing. The Pohang rate is also important, and yes, only the slope changes slightly, but it is in the direction of the slope gradually decreasing. It has dropped even further from the situation in 202.
In particular, regarding the Land market bit growth, the growth rate, which was about 30% before the launch of AI, has now dropped to 19% . So, the slope has gotten a bit lower. It can't be helped that it stays daytime. The parameter size is getting larger. Because the market is growing. Well, that could be the case .
So, in the end, actual aggregate demand did not increase much further. And memory cycles have nothing to do with demand in the first place . If we draw it together, hmm, [clears throat] 100% of memory cycles are caused by supply . Therefore, since this is caused by changes in supply, our analysis indicates that not only has demand not increased, but even if it had, it would not affect the memory cycle.
While demand continues to increase, the slope fluctuates; if you plot the overall trend, there may be slight spikes year by year, but if you plot the trend, it is a continuous downward trend. But now, yes, if we were to accept this as a fact, we would naturally assume that demand has been continuously increasing, and then there was an AI explosion in the last two or three years.
So, the question is, what is needed for AI? It is true that you need an Nvidia chip, and naturally, Nvidia chips require incredibly expensive high-bandwidth memory . Or this is right too. So, regarding Nvidia's performance, Nvidia was also a graphics card company before the launch of AI, right? Operating profit must have increased about 12 times since then.
That's right . yes. I sold the item for 12 times the price. yes. The problem for memory companies is that they sold the same amount of products but only the prices went up. Ah, it’s not that the demand for memory has increased . Production volume is the same as a year ago . Even regarding memory capacity in domestic semiconductor manufacturing, if a 1GB one used to cost 10,000 won in the past, a 1GB one is now 100,000 won, so sales and operating income are bound to increase .
So, in the case of NBD, the operating profit margin is in the 70-something to 80% range, making you wonder if it is really a company. Similarly, Samsung Electronics and SK Hynix also went up to around 80% . That's right . yes. So, isn't that not really strange ? When it comes to sustainability, the operating profit margins of a company that increased its profit by selling ten times more goods despite having nothing are completely different from those of a company that sold the same number of goods but saw its price rise tenfold.
Ah, so the price is bound to fall in proportion to what is absolutely right . But even if the increase in volume slows down a little, it won't take 10 minutes [clears throat] . The difference in sales volume is huge. ND would n't want to buy memory that expensively, and countless big tech companies feel the same way.
You wouldn't pay that money while getting ripped off so badly, would you? But considering that they are doing those things now and have been supplying them consistently, the fact that the market values them so highly is proof that there is demand and that they can be sold at a high price, isn't it?
Price is a matter of the balance of supply and demand . yes. If supply operates at 100% capacity but demand exceeds it, prices are raised. yes . If demand remains the same but supply decreases by 20%, the price goes up. So now, companies are reducing supply, so the supply has decreased significantly.
The total global output of Silko Weapons has decreased. Production itself dropped sharply by 20% from 2023 . The memory has been reduced. It's a line. We drastically reduced it and are gradually increasing it, but in terms of operating rate, it is only 70 to 80% . Factory operating rate? yes. This is data released by the government .
Is that not my point, but an unavoidable circumstance? Or, they could reduce the operating rate by exercising restraint in production. It's controlling it. The reason is that if companies operate at full capacity, prices drop immediately. So, for example, whether it’s Samsung Electronics, SK Hynix, or Micron, it would actually be very rare for just one company to have such high prices.
So, if they decide to quickly increase production capacity or raise factory operating days to make money, they could earn a tremendous amount . That is what Micron and Changshin Memory are doing . So, Samsung Electronics and SK Hynix aren't doing it. These two companies still have a 70% production cut rate to maintain prices, because their market points are the highest.
What do you think will happen the moment these two companies release their volume? It immediately leads to an oversupply. Even if Changshin Memory releases its supply, it won't have a significant impact because its market share is only 10% . Even Micron is much smaller than the combined total of Samsung Electronics and Hynix.
hmm. So, let's increase production. Even so, the price isn't falling, but since Samsung Hynix has a high market point, supply is bound to be released the moment they make a move. So, those people, Samsung Electronics and Hynix, are currently engaging in so-called collusion . Whether or not there was collusion is a matter for the court to decide .
That is not something I can presume to say, but it is a fact that production is being cut. So, if the two of them collude to cut production—because even if they just increased the mutual production volume from 80% to 100%, their profits would increase much more—but whether profits can actually increase is a question.
Ah, if the volume increases by 30% and the price drops by 30%, the operating profit margin will inevitably decline, right? yes . Even if the total amount of profit has increased, that itself becomes the trigger for the stock price to plummet. But if Samsung Electronics and Hynix intentionally... well, regarding the current price...
yes, I mean, you shouldn't talk about it so easily . Since it has n't been confirmed, I mean, if they did that, they would face a massive fine or something, would n't they? So, the lawsuit finally began in the California federal court in June. yes . Yes, targeting Samsung, SK Hynix, and Micron 3. Their memory companies intentionally reduced production of specific products to raise prices, so they are there .
By creating an artificial supply shortage, yes. A lawsuit was filed demanding triple compensation for the damages caused by the price hikes, and the important point is that this started as a class action lawsuit . The scary thing about class action lawsuits is that if just one party wins, many Big Tech companies that participated in the lawsuit can all suffer the same damages .
Ah, right. I don't know the result. It is currently in progress . That is currently in progress. Ah, so those companies are like, "Hey, this is a bit suspicious." Aren't you guys being too bold? Come to think of it, there are 15 people who filed the lawsuit, and they are individuals. Small and medium-sized PC companies.
Hmm. Even if these companies multiply, they wo n't receive much. yes. [Clears throat] The lawyer fees will cost much more . Right ? And they went against this giant corporation. yes. Is n't that strange? Ah, because such a reckless thing would leave a reckless thing behind. yes. From my perspective, I think this is basically Big Tech companies like Apple and Microsoft putting them forward as dummy targets to file a lawsuit.
If they happen to win, it would be enormous, but even if they lose, there isn't much to lose. Since they are only paying for the legal costs, if they win, they will reap astronomical benefits. yes . Unless you filed the lawsuit yourself or are the principal party, there is nothing you can gain. In a class action lawsuit, even if only one person wins, they receive the same compensation for damages as long as they can prove the same damages to the court.
So, regarding the day before yesterday, Apple said we also suffered damages. You sold it at a high price, so you just need to provide an explanation to the court. It is also not true that Big Tech companies currently need a massive amount of memory. [Snoring] It’s true. Yes, the fact that they are buying those things at a high price is actually because it’s the half-god they wanted.
So, well, that is why people sometimes openly raise complaints, right? If the complaint has reached the point of being made public, you have to consider how long they have been preparing to take action behind the scenes. hmm. So, right now people are buying it at incredibly high prices, but they suspect that this price is not based on natural supply and demand, but is artificial .
That's right . yes . To put it simply, it is a semiconductor boom created by reducing supply and raising prices. In a situation where prices are maintained even while controlling the supply volume, the economy doesn't really need to invest in equipment or facilities, yet they are doing so anyway . You're saying that you will continue to increase the supply, are n't you?
Those two actions are n't contradictory. Min is foreshadowing that future supply will intensify. huh . The supply oversupply is foreshadowing that future supply oversupply will become much more severe . The point of making such public facility investments is. But I can't not do it. You earned money, didn't you?
Anyway, this year. yes . There is no way to do it unless you invest. What does it mean to warn that it will end up being either labor costs or investment ? It's not that warning, yes. My assessment is that supply is already excessive, yet they are forcibly creating a supply imbalance. By reducing production.
But you know we keep building the strategy right now. yes . That means not only normalizing production, but also... Of course, in the future, it will be about supply, but they are building factories very aggressively. So, when talk of the Honam cluster comes up, the level of future supply and demand implies that the supply situation we are currently in will be intensified to a tremendous degree compared to what we have experienced in the past, and we are essentially seeing a preview of that.
The thing called investment. Then the price will drop drastically, so you're saying it will result in a deficit. Then why do companies make such decisions? Since I have the money, even if I don't invest, I have to pay taxes anyway. I couldn't help it . Uh, it’s something that absolutely has to be done.
And besides, the chicken game is about killing the opponent first and surviving, even if it means running at a loss. Yes, Samsung Electronics won the chicken game by continuing to do business at a loss . I intentionally lowered the price. So, by increasing the supply, the value is in deficit, but since the opponent's capital is running low, if they collapse first, you raise the store again .
hmm. But now, well, it’s not my fault. I'm not trying to nitpick, but aren't taxes something I have to pay anyway, regardless of the circumstances? Ah, but I suppose our country's law is quite old now. Yes, there was no tax on valid cash in the past. A very long time ago. However, currently, there is a tax on cash held that is not used for investment, labor costs, or R&D .
separately. So there is no need to keep it. yes. If you have it, you have to use the cash . You have to invest. Why keep that just to pay taxes? Couldn't you share some of that car with that shareholder? [Laughs] Have n't you been doing that for a long time? So now we have to pay off the shareholders, pay off the factory, and anyway, we have to use up the cash.
You have no choice but to get burned. yes. But regarding the "chicken game," for example, if about ten companies were vying for market share like this, it would be a game where the top two or three just wipe out the rest. But right now, it's Samsung Electronics, SK Hynix, and Micron, so who are you talking about playing a chicken game with again ?
So, in my view, if the three-company system falls within 3 or 4, I judge that the chicken-ke (chicken company) and Ganseong Hyeonjeong (political) levels will decrease . yes. Looking at past cases. yes. I believe there is still a possibility of a chicken dinner if it becomes 4-4-5-4, but even the Empress went bankrupt in Japan's LP in 2012, didn't she?
At the same time, the memory 44 system was maintained as a 3 system for 14 years, but now it has returned to a 44 system. Because it has exceeded a 10% market share in China. Changsin. yes. It's Changshin. Changsin. In that place, toothbrushing memory. The amount is higher. I heard there is talk of a 17% market share as well.
yes . This one already surpassed the LP market a long time ago . hmm. It has become a perfect private system. So now we've gone back to the days of the old Sickin' game . The sound that makes up the market . So, are you saying that companies need to prepare for a chicken game? Now, even if you do n't prepare, it is your fate to have no choice but to invest if you have money anyway .
Oh, really? So, the memory cycle is bound to continue forever; if it becomes compatible, you must definitely target it whenever you have the money. hmm. hmm. Then, in one or two years, you will inevitably have no choice but to become a doctor. yes . What we have been anticipating for months is that the scale of data centers will be completely different from what we currently know.
For example, if data centers are built on gigabyte scales, it means places the size of about 200 sports fields will be created. Just thinking about it, it feels like an astronomical amount of memory would go into that . But it is n't just happening in Korea; there are going to be massive AI data centers popping up in various places in the U.S., Korea, Japan, and pretty much everywhere else.
Also, if companies start establishing their own AI data centers, no matter how I think about it, it seems like memory demand will skyrocket. However, it does n't feel like it will just increase linearly, but rather make a complete jump, right? The reason this is impossible is, well, it's simple arithmetic.
Yes, if the current production volume is 100 units, yes, yes, if it increases by 100 units, it doubles. Yes, if the production volume reaches 10,000 units and increases by 100 units, um, it increases by 1%. Right? To give you an idea of the scale of the current production and shipment volume of D and Land, well, in the case of Land, it is 1.8 million Terrors.
Ah, it's petabytes. PETA is 1 trillion. yes . yes. It's one thousand trillion multiplied by 1.8 million. hmm. Because the market has grown so large currently, even if the annual increase is astronomical, the growth rate in the population cannot be high . No matter how much you listen, the denominator is too large.
That is why the growth curve of the industry cycle slows down as it moves into the market. It’s not that demand suddenly decreases; rather, even if demand increases, the market parameter grows larger, so the growth rate gradually slows down until it inevitably flattens out . Smartphones already peaked between 2012 and 2016 and have been on the decline, and automobiles also peaked in 2017 and are already decreasing; this is an inevitable fate that all industries must face.
So, regarding RAM DDR, that HBM is n't exactly a completely different market either. Then, in a similar market, HBM is a somewhat special case that uses only AI, so it is more vulnerable. It's more vulnerable. LPDDR can fit into HBM, mobile devices, and PCs, so it has no use in AI. yes. It is demand that has nowhere to go the moment the AI market's growth rate suddenly drops sharply.
I anticipate that the moment the growth rate slows down will be around next year, which is the expected timeline. I predict that the AI bubble will probably become a major issue in earnest sometime next year. The AI bubble? Because you lose the law. No matter who looks at it. Ah, I'm not saying that AI is wrong—that it will be the biggest, uh, big change in the future, big enough to almost completely alter the way of life for humanity.
yes. yes. So, the direction of AI development is also correct. It is true that the market continues to grow and various app gangs are emerging. yes. However, there is a speed at which the market grows . Demand, final demand— that is much faster. The problem is that the speed at which data is being invested to make money with AI is several times faster than that.
That is why it is a bubble. There is that discrepancy, and that discrepancy—the mismatch between supply and demand—is what constitutes the bubble. Uh, I did n't mention that the drop in memory prices is due to a decrease in demand . yes. It is falling because the supply is increasing faster, and the situation is exactly the same for AI right now.
Uh, because the reason the demand for AI can no longer increase at a higher rate is, well, first of all, [clears throat] there is no ultimate entity making money from AI. They say OpenAI makes the most money from AI, but it is actually operating at a loss. yes . There is nothing else to say about limiting AI.
The situation is such that there is almost no profit at all, and the problem is the supply side trying to make money from it. It is an AI data center supply. Yes, in the AI market. More than trillions of won have already been invested, and you're saying you'll do it again next year . But right now, there are no companies willing to pay to use that.
Ah, let me ask you a little bit about the technical aspects of this . As you mentioned, the bitrose has been steadily increasing all this time. But, well , right now in our country as well, as semiconductors become increasingly ultra-precise, the level of directness increases as more data is packed into the same size.
yes. But that is actually one thing that happened . yes. So, what happens now is that you will need more capacity anyway. If the limit of one per chip is exceeded, you have no choice but to increase the quantity. So, regarding demand, even if bit growth increases at a similar rate to what it has grown so far—that is, the number needed might be much higher than the current increase—than the current growth .
I'm not sure what you mean by "number," but capacity is what matters in terms of demand. Uh right, the capacity is bits. I wasn't saying anything special, I was just pointing out that the bit loss is gradually decreasing, so I'm not sure what you are imagining regarding the increase in the number. In the past, we could just attach one memory module to a single chip, and as the capacity gradually increased, everything was resolved.
But now, we’ve reached a situation where we have to cram 48, 100, or even 200 memory modules onto that main chip with just one GPU. The total amount below is the sum of all of that. Well, that is true. yes. I was saying that it is becoming more gradual. In the beginning. yes . So, regarding the idea that 100 bottles of 100ml are needed, yes, until now, one bottle was n't 100ml but increased by 1L, but since we are now in a situation where we can't increase it any further, it does n't mean that the number of bottles itself needs to increase.
When talking about price, it is hard to just think in terms of price per unit. It’s not like there can be a change in the total capacity. It does n't mean that the number becomes large just because it is split into several. Anyway, regarding the total volume, if that is the amount of alcohol you have to drink, splitting the vat doesn't change the total amount you drink .
yes . The total amount of beverages to be consumed will not be at a growth rate, right? That is bit growth, and while it is true that it continues to increase, as I mentioned, bit growth itself is bound to keep decreasing because the sample size is becoming too large. However, since the supply comes from China, new players are going to churn it out like crazy.
The point is that the rate of increase in supply is bound to be much faster . Because in the future, the U.S. will create China regardless of whether or not there is any containment of China . yes. yes. Then, if China stops importing from other places, the supply will eventually increase. Next, the important thing is that one of the things we have heard from many people is that Samsung Electronics still has the best technology and quality.
Of course it's good. It 's a leading company, you know. So, in the end, Chinese companies are for the domestic market. So, I think you have probably heard people say that competing with Korean companies is still a few years away . yes . Then, we would need to examine what percentage of global memory demand China accounts for .
hmm. It will be about 60%. It is a global city of the 60 Chinese market. So, the demand? No, regarding demand, ah, global buying... 60% of the demand for memory is China, and 60% of the total demand for semiconductors is also China. okay . So, Samsung Electronics is making memory to compete in China .
To sell in China. yes. But the very idea that there is no problem with replacing products made in China with Chinese ones is a bit of a miscalculation. Right ? You might think it’s not really about China. Another point of interest is whether there are any exports from China. Actually, I recently compared Korea's semiconductor exports with important semiconductor exports .
hmm. I found out a surprising fact. yes . Since 2019, China's total semiconductor exports have surpassed those of Korea. Exports are also to Korea. yes . Of course, since that includes foreign companies as well, it is not exclusively Chinese companies, and although the distinction is unclear, I noted that it has increased rapidly since 2019.
In 2019, during the Trump administration's first and second terms and the Biden administration's tenure, the semiconductor containment against China was extremely severe, so foreign companies did not issue certificates at all during that period . Hmm, good. In that case, it is highly likely that the increase in production and exports over the years is mostly due to volume from local Chinese companies .
Where do you export it to? I need to look at the data for that. I have a drawing. Because, aside from China, the place that uses these semiconductors the most is likely the United States, and the U.S. also sends a lot of people to Vietnam. Regarding Samsung Electronics' smartphone factory, hmm. Ah, so, regarding those factories in Vietnam, they are going to Vietnam and also to Taiwan.
Since I have n't looked at the figures again right now, if you are curious about what I have drawn up, I will show you a few countries right after this is over. Anyway, exports to China have now surpassed our country's total amount, and we haven't been able to overtake them even once in the last five years .
Oh, really? You'll have the data soon, after all . yes . And now, we have to view it as impossible to turn around . Because production continues to increase while Korea is reducing it. Looking at the situation right now where the U.S. is restraining it, if that loosens up a bit, much more could come out.
Once the domestic market fills up, they have no choice but to push out products, right? However, for the time being, domestic demand alone is sufficient, so they might not need to worry too much about exports . yes . Now, assuming that we fully accept such things, nevertheless, Samsung Electronics' quarterly profit was, for example, 100 trillion won and 107 trillion won.
yes. We paid 107 trillion, and if we include this year, it's probably around 300 trillion. I think it will be about that much. Even Hynix [sneers] well, anyway, they are generating record-breaking, truly unimaginable profits, so even if the future outlook for those companies is a bit bleak, there is an assessment that their current prices are actually very cheap, you know?
So, actually, since I am not a stock market analyst, discussing whether stock prices are cheap or expensive is, in a way, a byproduct. Yes, that is. yes. My analysis is whether memory prices will rise further or fall . yes . Memory prices are still rising, but yes. I believe the trend will shift to a decline as early as the fourth quarter or at the latest by the first quarter of next year.
Regarding memory prices, regardless of the type of memory you are referring to, I am not sure about HBM prices as there are no statistics available and I do not have any myself. However, since HBM accounts for only 6% of the total memory market, it does not hold much significance. Ah, but that is true.
However, since HBM is the most important to our companies, HBM probably only accounts for 10% of total production. Since they are busy making HBM, they can't produce DDR anymore, so there is an aspect where DDR prices have risen quite a bit . no. Then only companies do all the talking . No one has seen any data to prove that .
We now understand. If that were true, they wouldn't have filed a lawsuit. They filed a lawsuit in the U.S. because they were convinced it was not true . That is correct. You're convinced it's because production was reduced, right? So now, memory prices will also... starting this fourth quarter at the earliest, or the first quarter of next year at the latest.
Now, if memory prices drop, they are bound to fall very quickly . It's falling because of an oversupply, is n't it? yes. As I mentioned earlier regarding the price, it is very rare for the balance to be perfectly maintained and last a long time. Ah, it's either insufficient or excessive. I'm always going back and forth.
yes. So, it is bound to rise steeply and fall steeply, but once it starts to fall, the price [snorles] is bound to drop to a normal price . That is the price prior to April 2025, and since Samsung Electronics and Hynas's stock prices are 70,000 won and 200,000 won at that time, I am saying, as a byproduct, that the stock price will naturally go up to that point.
yes . If by any chance [snort] Electronicics um [clears throat] did something like that, I will sue too. I'm also right now, yes. So, since we cannot predict when a lawsuit will be in progress, is it correct to say for now that the litigation is ongoing ? So, the fair price of memory that you see, Doctor, is roughly the price from around April of last year, and even before April, it had already gone up due to a year of production cuts .
So, at that point in time, the stock prices are around 70,000 won for Samsung Electronics and 200,000 won for SK Hynix because profits are bound to return exactly as they are. These are the right stock prices due to the current profits. Around when? I don't know if that will happen in a year or a year and a half, as I don't know the speed of the decline, so it will depend on the speed of memory price drops.
However, since stock prices are likely to bottom out before memory prices do, they might rise a little sooner. Therefore, the exact timing is unknown right now, but once memory prices start falling in earnest, we will be able to make a prediction to some extent by observing the slope of the decline . yes.
Comments 0