YT YouTube2Text

YouTube2Text → Transcripts → [주식] 261016 이제 이 종목은 보고있어야 합니다

[주식] 261016 이제 이 종목은 보고있어야 합니다 — Transcript

by 한균수의 주식사용설명서 · 7,460 words · 1,124 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:03Hello. It's Hansoo again today. You've
0:05all worked hard this week. It’s
0:06Friday, and looking back at this week,
0:09it’s truly been a disaster. There
0:11have been so many absurd days while
0:13watching the market that I actually
0:15said something yesterday. When the
0:17plunge happened, everything hit the
0:18circuit breakers. Is the capital market
0:21trying to perform a format or a reset,
0:23like we do with computers? I'll look at
0:26the KOSPI index chart in a moment, but
0:29I once wrote a post about this. It was
0:32July 29th. Remember that day. That day
0:34was actually the first time during a
0:36public broadcast where I spent the
0:38whole time just venting in frustration,
0:39and the KOSPI has now nearly returned
0:41to the levels of that day—it makes no
0:43sense. Of course, the KOSDAQ still has
0:45some room to fall further, but the fact
0:47that we've even reached this point is
0:48absurd. If you look at Samsung and
0:51Hynix, there’s a lot of overhead
0:54supply from people who bought in at
0:56higher prices, creating a burden on
0:58supply and demand, but their forward P/
1:01E ratios have fallen to the low 3x
1:03range. It’s ridiculous. How could
1:05this happen? I could understand if the
1:09cycle was ending immediately, peaking
1:11this year or the first half of next
1:13year, but even though I'm not someone
1:15pushing to buy IT stocks right now, the
1:17cycle won't end until at least 2028.
1:20Yet, with a forward P/E of three, the
1:22fact that these stocks are still dying
1:24is just nonsensical. I think we’ll
1:27remember this year, 2026, even 10 or 20
1:29years from now; as you all know, I’ve
1:31been in the stock market for quite a
1:34while, and I’ve seen so many
1:36unprecedented things this year. The
1:38problem is that it’s only happening
1:39in our country. Looking at the recent
1:41drops in Asian and U.S. markets, yes,
1:43Asian markets were generally poor
1:45yesterday, but we are falling
1:46significantly more. And yet, there is
1:48no specific bad news for us
1:49individually. Foreign investors were
1:52buying Hynix ADRs because its market
1:55share is higher than Micron's, and many
1:57were pouring into our market, but now
2:00they are throwing their hands up and
2:02leaving. Can you believe a market like
2:04this exists? It’s showing such an
2:07exceptional and unusual trend. I’ll
2:09stop here because I need to quickly get
2:11into what I’ve prepared for today, so
2:12let’s get started. The KOSPI is up 2%
2:14today and the KOSDAQ is up 4.46%, so
2:16the KOSDAQ rose much more. But the
2:18KOSPI has 759 advancing stocks. If you
2:21look at yesterday's market, the
2:22composite index dropped by 7-8%, and
2:25when the index falls by that much, you
2:27have to assume that virtually every
2:29stock is falling. Nearly 98%of stocks
2:31were just moving in one direction
2:33together, but when you break it down
2:35like that, it means there’s the
2:37remaining 2%that actually went up.
2:40There was a very important stock among
2:42them. So, I'll talk about that stock a
2:44little later, and starting next week,
2:45that stock will be the benchmark.
2:47We’ll look at that later. And if you
2:49look at today's market, it fell a lot
2:50yesterday. When it tried to rebound
2:52today, most stocks went up; in the
2:54KOSDAQ, there were nearly 1,500
2:55advancing stocks and only 200 declining
2:57ones, so you can see that most things
2:59were doing okay. And while everything
3:01in the KOSDAQ was good today, the
3:02biotech sector was particularly strong.
3:04It was very strong. But there was a
3:06reason why it had to go up. And adding
3:08to that, if we talk about
3:09beauty-related stocks, things like ECM,
3:11Skin Booster, LNC Bio, and Hans Bio—
3:13they really took off today. So, those
3:15kinds of stocks rose quite strongly,
3:16and most others, like materials, parts,
3:18and equipment, were also fine, while in
3:19the KOSPI, defense stocks moved a bit
3:21more strongly in the morning, which
3:22also had its own catalyst. In this
3:24sector. So, in any case, most of them
3:26were at least rebounding. But looking
3:28at today's market, the thought that
3:30came to mind, if I had to summarize it
3:31in one word, is that it's a kind of "
3:33ethical rebound." If they have a
3:35conscience, they should at least
3:36rebound by this much, and there are no
3:38holidays until Christmas for the time
3:39being. Excluding the weekends. Since
3:42there are many trading days left, next
3:44week’s five trading days should
3:46confirm that the gap is filled by a
3:47V-shaped recovery. It took exactly four
3:49trading days for the index to break
3:51this significantly. It fell too much
3:52for four consecutive days, which is why
3:54it came up here today; if we add
3:56another three or four trading days here
3:57, a V-shape will form, and since I
3:59believe it will form, I think we can
4:01just use that as a benchmark for next
4:02week. Now, let's look at the composite
4:05index with our eyes. This is KOSPI, and
4:07this is KOSDAQ. As I said, it melted
4:08down quite hard over about four trading
4:10days; here's one, two, three, four, and
4:12it had actually fallen once more during
4:13the five trading days before that. So,
4:15it didn't take that long for it to come
4:17down this much. The KOSPI composite
4:20index fell 22%. If you draw from the
4:22top here, the composite index is down
4:24about 18-19%. KOSPI fell more. A little
4:27more than the KOSDAQ. And the rebound
4:28seen today was also larger in the
4:30KOSDAQ than in the KOSPI. And if you
4:31look at the days when the index was
4:33strong, over the past one or two months
4:36, there were far more days when KOSDAQ
4:38large-caps rose more strongly than
4:40KOSPI large-caps. So, in the big
4:42picture, a rotation from large-caps to
4:44small and mid-caps was being confirmed,
4:46but it was going well and then it died
4:48here; ultimately, what I'm talking
4:49about is this 800-point range in the
4:51mid-to-high levels, right? It’s at
4:54794 points now, and it shouldn't take
4:56long to get up to here. If we take it
4:58easy until around Thursday or Friday of
5:00next week, reaching about 880 or 890
5:02would make it perfect. I think it's
5:04possible, but anyway, I’m telling you
5:06to use this as a benchmark, and
5:08although there was a rebound today, the
5:10intraday low for KOSPI yesterday was -9
5:13%. 9. The KOSDAQ low was also -9.5%. So
5:16it dropped incredibly hard yesterday,
5:17and you probably all know there are
5:19three major reasons why it fell so much
5:21. The first is that the corporate bond
5:23prices of CoreWeave have fallen to the
5:25level of distressed debt prices. The
5:27second is that since Anthropic is
5:29facing an IPO, they had to pump up
5:31their company's stock value, so they
5:32came out to say their quarterly revenue
5:34was decent. They came out to say they
5:37did well, but there were some question
5:38marks regarding Anthropic’s revenue.
5:41That became one of the reasons for the
5:43drop, and the third and more important
5:45thing was the report content from a
5:46market research firm. Because of this,
5:49it fell quite a bit. We need to know
5:51why it fell, focusing on those three
5:53things, so let's look at them one by
5:54one. Before we look at our country,
5:56let's look at the US stock market from
5:58last night and the night before. You
6:00see how US semiconductors were melting
6:01down the night before last, right?
6:02Meanwhile, biotech stocks took off. And
6:04if you look at the market yesterday, US
6:06semiconductors weren't very good
6:07yesterday either. Of course, Micron
6:09closed up 1.87%, but early in the US
6:12session yesterday, Micron was down by
6:14as much as -6%. So, seeing it flip to
6:18positive and rise to plus 1-2%, I
6:20thought last night and this morning
6:22that "Ah, it won't die tomorrow," but
6:24the rebound was actually quite weak.
6:27It’s certain that the effect of the
6:28previous day's news has worn off
6:29significantly. And the sector that was
6:30strong again yesterday was biotech. So,
6:32there are actually more US biotech
6:33stocks that are stronger than you might
6:35think. Let’s look at those a bit
6:37later, and for now, let’s go through
6:39the details. First, both CPI and PPI
6:41have been released. Looking at the CPI
6:43announcement, the forecast was 0.2%,
6:45but it came in at 0.3%, and it was 3.6%
6:47but came in at 3.7%, so the CPI was
6:49higher than expected. Then, naturally,
6:51the market should be bad, right? It
6:53wasn't good for the market, but
6:55ironically, even though the CPI spiked,
6:57bond yields plummeted. This played the
6:59biggest role in biotech rising. The
7:01speed at which bond yields are coming
7:03down is incredibly steep, beyond what
7:04we expected. This is the 30-year bond;
7:05this shows the drop from the night
7:06before last, and this is the drop from
7:08last night. So, looking at it based on
7:10the 30-year bond, just two or three
7:12days ago it was breaking through 5.6%,
7:14but now it’s at 5.3%. Is it a
7:17significant drop if it falls by 0.2? It
7:19is an enormous drop. In fact, if it
7:21goes down in this structure, a massive
7:23surge is inevitable. Always the problem
7:25. So, oil prices weren't jumping—well
7:27, they have been rising recently—but
7:29with oil being stable, moving sideways
7:31in the 90-dollar range, most indicators
7:33looking good, and the U.S. stock market
7:35performing well, the only thing I was
7:37worried about was bond yields silently
7:39rising every day; but if they drop like
7:41this, growth stocks have no choice but
7:43to rise. Actually, there wasn't any
7:46major news for U.S. biotech. But since
7:47biotech benefits the most when bond
7:49yields drop significantly, healthcare
7:51and biotech were strong yesterday and
7:52the day before as well. So, let's look
7:55at the charts of some large-cap U.S.
7:57biotech and pharmaceutical companies
7:59before moving on. We can't skip looking
8:01at Lilly, as a representative example.
8:03Yes. It was strong yesterday and the
8:05day before. And adding to that, Merck
8:07...It plays the role of a leader in
8:11anticancer drugs. Or Johnson&Johnson;
8:13these are the big players among U.S.-
8:14listed biotech and healthcare companies
8:16. So, take a look. Lilly has already
8:18reached near its all-time high. And for
8:20Merck, there isn't much left, while Eli
8:22Lilly broke its all-time high yesterday
8:23. And Johnson&Johnson isn't falling
8:25much from the bottom either. The
8:26important thing about these new highs
8:27is that these are companies that
8:28already hit new highs back in August
8:30and September. They've already broken
8:31through. So, looking at this, our
8:33domestic biotech companies should
8:34naturally follow, and aside from them,
8:36at least Celltrion and Samsung
8:38Biologics should be held; when the
8:39KOSPI circuit breaker hit yesterday,
8:41Celltrion and Samsung Biologics were
8:43actually rising. So, I thought, "This
8:46could work," yesterday; you can see
8:48that biotech stocks are the ones most
8:50affected by base interest rates, and
8:52the first important issue is this.
8:54CoreWeave corporate bond prices; you
8:55know what kind of company CoreWeave is,
8:57right? It is a representative GPU cloud
8:59company. The original business model of
9:02a GPU cloud company is, well, GPUs,
9:04right? They buy NVIDIA GPUs. Then they
9:06plug them into their own data centers.
9:08So, they don't use this computing
9:10infrastructure themselves; they run it
9:12like a rental business, much like
9:13someone buying apartments to lease them
9:15out. To their customers. But since they
9:17eventually need to buy a lot of GPUs,
9:19they don't have the money for it. These
9:20cloud companies. So, they raise funds
9:22externally, like through paid-in
9:24capital increases. Then, who is the
9:26company that typically provides a lot
9:27of money to CoreWeave? It was Nvidia,
9:29right? The structure is quite unique,
9:31which is why the term "circular economy
9:33" came up; CoreWeave needs GPUs but
9:35lacks the funds. So, Nvidia lends them
9:38the money. They either invest or lend
9:39it to them. Then CoreWeave uses the
9:41money received from Nvidia to buy GPUs
9:43from Nvidia. This then shows up as
9:45revenue. Using this as a basis, they
9:47raise more funds or use the money
9:49they’ve made—though they’re
9:51mostly running at a loss—to buy even
9:53more GPUs. It’s as if Nvidia is
9:56creating its own demand. To put it
9:59simply, this structure is exactly like,
10:02for example, me moving into a new home
10:04for my honeymoon; I just set up my new
10:07married life. I’m moving in, and I
10:09have a refrigerator I’ve been using
10:11for about 10 years. I intended to keep
10:12using it, and while I’d love a new
10:14fridge, I don't have the money. But
10:15then Samsung Electronics comes to me
10:17and says, "We'll lend you the money, so
10:18why don't you use it to buy one of our
10:19fridges?" So I bought the fridge after
10:21they told me to pay them back later.
10:23Even though I had the desire to buy one
10:24, you couldn't really call me a "
10:26genuine consumer," right? Why? Because
10:28I wanted one, but I was in a state
10:29where I couldn't buy it because I had
10:30no money. But Samsung is saying, "We'll
10:31give you the money, pay us back later,"
10:33effectively telling me to buy their
10:34fridge with the money they provided.
10:35That’s closer to creating demand out
10:37of thin air. That’s why the talk of a
10:39"circular economy" emerged—the idea
10:40that they are squeezing out demand
10:41where there isn't any. Nvidia has this
10:43kind of structure, and they already own
10:46about 10%of CoreWeave's shares. Nvidia
10:48is one of CoreWeave's major
10:49shareholders. But they issue a lot of
10:51corporate bonds. Because they need to
10:53buy GPUs but don't have the cash. When
10:55the interest rate—or yield—on these
10:57bonds goes up, it means the price of
10:59the bonds is falling and they aren't
11:00selling well. From an investor's
11:04perspective, the reason they don't find
11:06merit in CoreWeave's corporate bonds is
11:07that while they understand the business
11:09model of raising capital to buy GPUs,
11:11you need to eventually generate revenue
11:13and results from leasing out that
11:15computing infrastructure to customers
11:17to pay back the debt. The market is
11:20starting to think that you won't be
11:21able to pay it back. You can just buy
11:23government bonds and get a 5%return,
11:25which is basically risk-free income.
11:27It's guaranteed unless the country
11:29collapses. So the market thinks, "If
11:31you're offering such low yields, why
11:32would I buy your bonds instead of
11:34government bonds?" The yield on their
11:37corporate bonds actually climbed to 15%
11:38. But when the yield to maturity hits
11:41around 15%per year, it essentially
11:43means it's a distressed bond. Once the
11:45band for distressed debt crosses 15%,
11:46they are all considered bad debts. In
11:48other words, the market has started to
11:50interpret CoreWeave’s situation as
11:53dire, viewing it as unsustainable. The
11:55problem is that Nvidia is a major
11:57shareholder in CoreWeave. And Nvidia
11:58has plenty of cash. There’s been talk
12:01about whether Nvidia would bail out
12:02CoreWeave, but the situation is quite
12:04nuanced. Let's take a look at the
12:06business model between CoreWeave and
12:07Nvidia. It operates in this specific
12:09structure. A circular economy of sorts.
12:10If CoreWeave gets into trouble,
12:12Nvidia's equity investment would be
12:13compromised. CoreWeave was a huge buyer
12:15of their GPU supply, and they would
12:16stop buying. So it's difficult for
12:18Nvidia to either help them or just let
12:19them be. Let’s assume for a moment
12:21that they do help. Nvidia could easily
12:23bail out CoreWeave if they wanted to.
12:25But if they do, it creates a precedent,
12:28much like a bailout by the IMF. After
12:32all, CoreWeave isn't the only player in
12:33the U.S. GPU cloud market. Others are
12:35struggling right now, too. They would
12:36ask, "Why help them and not us?" and if
12:39Nvidia bails them out once, can they
12:41really stop next time? If they fail
12:43again, Nvidia would be expected to step
12:44in once more. Meaning they would have
12:46to do the same for other companies too.
12:48It creates a precedent for systemic
12:49bailouts. And if Nvidia does intervene,
12:51what would the market think? "Wow, they
12:54manufactured demand, and now that the
12:56buyer is broke, they're pouring in more
12:58money to force that demand.""So they
13:01really did create the demand themselves
13:02.""This wasn't a real customer after
13:03all." That is definitely how people
13:04would see it. So, it’s not easy for
13:06Nvidia to step in. But if they don't,
13:07and the company goes under, they lose
13:09all their money. Nvidia was caught in a
13:11dilemma, but CEO Jensen Huang spoke
13:14about it yesterday. What did he say?
13:17The best outcome is to resolve this
13:18without getting their hands dirty. He
13:20stated that since CoreWeave's financial
13:22health is improving, he won't be
13:24providing a bailout. Following these
13:25comments, CoreWeave saw a rebound
13:27yesterday. It had plummeted
13:28significantly the day before. That was
13:29the first reason for the IT sector's
13:31decline. And the second reason came
13:33from a report by the same market
13:35research firm, which stated that it
13:37would be difficult to recover the money
13:39invested in AI data centers. But this
13:41story sounds a bit familiar, doesn't it
13:42? We heard the same thing last year and
13:44the year before that as well. For
13:46reference, it was around the same time
13:48last year in 2023. In September 2023,
13:50there was some talk that AI-related
13:52stocks were similar to a bubble. The
13:54logic back then was nearly the same.
13:56Back then, the market didn't even react
13:58, but now that it's been dropping
13:59recently, people are saying, "That's
14:01right." The reason the market is
14:02pulling back is that people are
14:03thinking, "Now that I look at it, it
14:04really seems that way." An enormous
14:06amount of money is pouring into
14:07artificial intelligence. This year
14:08alone, AI infrastructure investment is
14:10known to be around $ 1 trillion, and an
14:12additional $ 6 trillion is expected to
14:14be invested by 2030. This means that
14:16based on forecasts, approximately $ 7
14:18trillion will be invested in AI by 2030
14:21. More investment is a good thing.
14:23Buying more semiconductors is all fine
14:25and good, but the problem is that the
14:27money invested here must eventually be
14:28monetized, and it isn't being monetized
14:30. So, for the investors who put money
14:34in, they need to see at least a certain
14:36level of revenue and performance to
14:38recoup their investments; they believe
14:41an ROIC (Return on Invested Capital) of
14:43about 12%is needed by 2030. To achieve
14:47that, cumulative revenue from pure AI
14:49would need to reach $ 8.5 trillion, but
14:52projections show that it will barely
14:54reach $ 3.6 trillion by 2030 at best.
14:57What does that mean? It means the money
14:59poured into this is evaporating. They
15:00are saying it won't be recovered. And
15:02they said something even scarier, which
15:04CoreWeave was heavily affected by. That
15:06over 80%of Neo-Cloud providers will
15:08cease independent operations by 2030.
15:12Simply put, this means the Neo-Cloud
15:13business model will disappear before
15:152030 arrives. That it won't work. Then
15:18what is meant by "ceasing independent
15:20operations" here? As mentioned earlier,
15:23companies like CoreWeave, which are
15:24Neo-Cloud firms, don't have their own
15:26money to fund this system. They raise
15:29external capital to buy GPUs and rent
15:31out computing infrastructure, so think
15:34of it like someone taking out a full
15:36loan at 3%interest to buy an apartment.
15:39If the rental yield is 5%, you earn a 2
15:42%profit, right? But interest rates are
15:44rising, and the cost of capital is
15:45going up right now. Or, if customers
15:47decrease later or apartment prices drop
15:50, rental yields won't be 5%, but will
15:52fall to 4%or 3%. Then, neo-cloud
15:53companies will just go bankrupt
15:55immediately. There is no way around it.
15:57It’s not that AI itself is ending.
16:00It's just that neo-cloud business
16:02models are so fragile that, like real
16:04estate, if the rental yield drops, they
16:06just collapse. So these guys will fail,
16:10and then who will survive? Big Tech
16:12will. Why? Big Tech companies already
16:14have a lot of money from their core
16:15businesses. And their cost of capital
16:17is not very high. Their bond interest
16:18rates might be high, but they are still
16:20much cheaper than neo-cloud companies.
16:22So, because they can raise funds at a
16:24lower cost, they will survive and
16:26probably absorb the others. In the end,
16:30it means there’s no answer. But the
16:31problem is that there was similar talk
16:33last year, yet it didn't really affect
16:35the market back then. And the third
16:37issue was this. Ahead of its IPO,
16:40Anthropic disclosed its revenue for the
16:42first three quarters of this year to
16:43investors, claiming a profit on an
16:45adjusted basis for the third quarter.
16:47Wait, I just said AI isn't making money
16:49, so how is there a profit? We need to
16:51dig into this. That's not a real profit
16:52. If we look at the details, it shows
16:57right here. Anthropic’s quarterly
16:59earnings report stated $ 4.7 billion
17:01for Q1, $ 11.6 billion for Q2, and $
17:0416.8 billion for Q3. That means it grew
17:06147%and then 45%compared to the
17:08previous quarter. The important thing
17:10is that 45%growth year-over-year is
17:12actually a huge amount. In terms of
17:14pure numbers. But to say that growth of
17:17only 45%in one quarter is a failure.
17:21The reason for this is that it's the
17:22same for semiconductors; the growth
17:24rate of their operating profit goes up
17:25until Q3 of this year, and then it
17:26starts to dip. It doesn't crash
17:29downwards, but the slope becomes
17:31flatter. We entered this phase starting
17:34from Q3 of this year. Of course, I
17:36expect it to hold for about two years,
17:38but it won't crash while at the same
17:40time it won't go up either. So,
17:42comparing it to when it was rising, why
17:44did it only grow 45%this quarter when
17:47it more than doubled in previous
17:49quarters? It’s because Anthropic had
17:52some issues in the third quarter. What
17:55kind of issues? It can be seen as this.
17:58On July 9th, their competitor OpenAI
18:00released GPT-4o, and as soon as it was
18:02released, we ran benchmark performance
18:04tests and found that it was better than
18:06Claude Opus. So, people who were using
18:09Anthropic are suddenly switching over
18:12to OpenAI, and I've been really
18:14praising GPT-o 1; this o1 came out on
18:16September 3rd. So, ever since it came
18:19out, I’ve been singing its praises
18:20nonstop, saying it’s absolutely
18:22insane. Oh, by the way, I'm going to
18:24take a quick tangent. I need to mention
18:26this because I forgot earlier: I used
18:29GPT-o 1 to code and build my own
18:31website. I told you I was working on it
18:32, right? Building it took some time,
18:35especially considering all the
18:37sleepless nights, but it was finished
18:39in just about a week or two. You can go
18:42to haninvest.co.kr or just search my
18:45name, Han Kyun-soo, on Naver; the
18:47content is honestly really good, even
18:50by my standards. If you search it,
18:52there’s a link for my official
18:53website at the bottom. If you click
18:54that, it takes you right there. Anyway,
18:57once you're in, I’ve coded a briefing
18:59that gets posted around 7:00 AM. So,
19:02it’s the actual content that I look
19:04at myself. I start every morning by
19:06reviewing this. It summarizes
19:07everything for me. It covers how the
19:09KOSPI index was yesterday, how the US
19:11indexes performed overnight, and I’ve
19:13coded it to display the S&P 500 map,
19:14which I always talk about—I’m
19:16constantly refining the prompts, so
19:17it’s getting more complete, and sure
19:19enough, it pops right up. Since I don't
19:22do live broadcasts every day, I really
19:23hope you all check this out. And it
19:25shows which stocks were noteworthy,
19:27like Dell dropping 6.8%, and explains
19:29why it fell. It also covers what
19:31happened and which stocks were strong,
19:33like Salesforce. AI software companies
19:35like them were quite strong. It
19:37explains that long-term contracts can
19:38lead to recurring revenue for AI, and
19:40if you look further down, I’ve
19:41categorized everything by major news,
19:42pros, cons, and even importance levels.
19:45I’ve refined the prompts so much that
19:47the quality has really improved; at
19:48this level, I think it’s pretty much
19:50a complete briefing. It contains all
19:52the information I need. Anyway, this
19:54morning's briefing is about whether the
19:55US market rebound will spread following
19:57the 7%drop in the Korean market. Since
19:58the US rebounded yesterday, the
20:00important thing in the US market
20:01yesterday was the PPI. The PPI
20:03discussion pops up right there. It says
20:04producer prices surged. So, it all
20:07connects perfectly, including TSMC's
20:09earnings report from yesterday; the
20:10briefing gets posted there, and it goes
20:12up every single morning. Then, around
20:14lunch or in the afternoon, there are
20:16the hot reports of the day, right?
20:17Dozens or even hundreds of reports come
20:19out every day. Even I can't read every
20:21single one of them. But honestly, when
20:24I look at reports, they often contain
20:26the same old information, so I can
20:27usually skim through them in two or
20:29three minutes. However, if there's new
20:33or updated content, I have to check
20:35everything, which takes a long time. I
20:37can't read everything, right? So, I
20:39pick out just three reports that had
20:41the most views that day or were the
20:43hottest topics in the market, providing
20:45links to the original sources along
20:46with summaries. I do this every morning
20:50and afternoon because the "hot" reports
20:52in the morning are different from those
20:53in the afternoon. I've set it up so
20:56those get posted, and as I was
20:58explaining before I moved on, the same
21:00statement—this one here, I wrote this
21:02myself. After I wrote it, I ran it
21:04through a prompt with AI, and this is
21:07actually the most important content.
21:09I'm trying to make it so that key
21:11economic schedules and indices are
21:12linked in real-time, but when I click
21:14this, it doesn't go anywhere, right?
21:16Why doesn't it go through? That feature
21:18hasn't been implemented yet. I have to
21:20link an API or something, but it's
21:22quite difficult, so I'm working on it,
21:24but it might take some time. Anyway,
21:26like that, this was made with Astra. I
21:29coded this with Astra Vibe, but
21:30regardless, once this came out,
21:32starting from the third quarter, while
21:33Anthropic’s performance is still
21:35increasing, the growth numbers have
21:36flattened out. So, people thought it
21:39would dominate everything on its own.
21:42While it’s still the top player in
21:44the B2B market, OpenAI's performance
21:46improved faster than expected, which is
21:48why growth is seen as slowing down. And
21:51what’s important is that they clearly
21:52stated they achieved an adjusted profit
21:54based on the third quarter. There’s
21:56no way this is actually profitable. So,
21:58if you look at the reported content,
22:00just look here. For example, regarding
22:03AI, OpenAI runs ChatGPT, right? Just
22:05think simply about the revenue that
22:08company brings in. How would they make
22:10money? Subscription fees from people
22:11like me. I pay hundreds of thousands of
22:13won a month for various services, so I
22:14use it a lot. So, subscription fees
22:16would be a typical example. That’s
22:18where the revenue comes from. Then, it
22:19has expenses or equipment investments,
22:21so where do those costs primarily go?
22:23For instance, server maintenance costs,
22:26employee salaries, and if there are
22:28stocks, stock options too—these are
22:30all expenses. And the important thing
22:32is, this is what costs the most. If you
22:34look here, broadly speaking, there are
22:36AI inference costs and training costs.
22:39Since you have to process a lot of data
22:41to get good results, training costs are
22:43incredibly high. Inference generates
22:46answers based on what has already been
22:48learned, so inference costs are lower
22:49than training costs. So, where do most
22:52of the costs come from? It’s not
22:53really from things like labor costs.
22:55They mostly come from here. From the
22:57training costs. So, no matter how much
22:58revenue comes in, the training costs
23:00are several times that amount. That’s
23:02why, structurally, it’s impossible
23:03for this to turn a profit. But I looked
23:05into what they manipulated to show a
23:06profit, and I broke it all down. It
23:08turns out they included inference costs
23:09in their adjusted earnings this time.
23:11They’re saying, "Inference costs
23:12aren't much, so it’s okay to include
23:13them." Employee salaries and general
23:14operating expenses don't amount to much
23:15compared to the staff. They say it's
23:16generally included, considering the
23:18scale of the company. But the important
23:20thing is that the AI model's training
23:22costs are reportedly excluded. So, of
23:24course, it would show a profit. It
23:25seems like they’re pumping up the
23:28numbers to tell a good story ahead of
23:31their IPO. I mean, when talking about
23:33company performance or profits, they
23:35just say, "Hey, just exclude this cost
23:36and call it adjusted earnings. What did
23:39we spend a lot on this month? We had a
23:41lot of staff dinners, so let's exclude
23:43those, call it adjusted earnings, and
23:44after excluding the big expenses and
23:46keeping the small ones, just make the
23:48numbers fit. Oh, it’s a profit? Then
23:50announce it as a profit." That’s how
23:52it’s going. Of course, you couldn't
23:54do this after going public, but since
23:55they're still private, these accounting
23:57standards are a bit ambiguous. Anyway,
23:59they claimed a profit to pump the value
24:00, but it's obviously not a real profit.
24:02If you add everything up, they are
24:03currently facing a massive deficit. But
24:05it’s not strange for them to be in
24:06the red; it’s actually expected.
24:07Anyway, those three things were the
24:09main issues. TSMC’s earnings came out
24:12very well. The PPI also exceeded
24:14expectations, but as soon as it was
24:16announced, bond yields fell and tech
24:18stocks rose. So, please just take a
24:20look at the earnings report yourself.
24:22And President Trump mentioned yesterday
24:24that he is considering dropping a 10-
24:25ton super-sized bomb on Iran. There has
24:28been talk that the U.S. might use nukes
24:30against Iran, not strategic ones, but
24:32tactical nukes. For reference, a
24:34strategic nuke is meant to wipe out an
24:36entire country, while a tactical nuke
24:38is for hitting specific, targeted areas
24:41. During the war in Ukraine, there was
24:42also a lot of talk about whether Russia
24:43would use tactical nukes or not. So
24:45people asked if they were going to use
24:46a nuke, and it turns out it’s not a
24:48nuke, but a conventional bomb on par
24:49with one. What was it called? They
24:51intend to use a super-sized
24:52conventional bomb. This story came out
24:53on CNN, and it caused oil prices to
24:55rise a bit. Please just take this as a
24:57reference. I already talked about
24:59Jensen Huang earlier, and there really
25:01wasn't any new news in the domestic
25:02stock market today. But if I had to
25:04mention just one thing, there is one.
25:06Everyone was talking about this all day
25:07long. I watched it live this morning,
25:09and it was so fascinating that I showed
25:12it to you. A car that actually flies.
25:14It really flies. At first, I thought it
25:16was made with CG and AI, but it was
25:18real. They said they would mass-produce
25:20it starting in 2028, and look at this
25:22car. They actually did a live demo of
25:24this earlier. Tesla just unveiled a
25:26supercar called the Roadster, which is
25:29an electric supercar. They unveiled it
25:32with plans to mass-produce it the year
25:34after next, and its biggest feature is
25:35that the car flies. It's a flying car.
25:38You know the technology SpaceX uses to
25:41control the rocket's attitude when it
25:43lands? The way the thrusters puff smoke
25:47to stabilize it before it lands right
25:50on the mechazilla? They supposedly
25:53equipped this car with about 12 of
25:55those thrusters. So they say it makes
25:58the car fly, and you should take a look
25:59at this. This is a true story. This
26:01isn't AI or CG. Look at the car. This
26:05was unveiled earlier. This is a real
26:07car. There is nobody inside right now.
26:09And look at this. Let's assume there's
26:11traffic congestion with cars ahead. The
26:14car flies. Do you see it? They put that
26:16rocket propulsion and attitude control
26:18technology into it. For reference, it
26:21says it goes from 0 to 100 km/h in 0.8
26:23seconds. Supercars like Lamborghinis
26:25and Ferraris are usually in the 3-
26:26second range. No, they said it's 0.8 to
26:280.9 seconds. I looked into how that's
26:31possible, and they put those thrusters
26:33in the back to push it forward when
26:35accelerating rapidly. That's crazy.
26:38It's absolutely insane technology.
26:39Because of this news, Tesla is
26:41currently moving up. It's up about 3-4%
26:43because something so shocking was
26:45revealed, so anyway, that was pretty
26:46much the talk of the day. And as for
26:48the domestic stock market, there's
26:50nothing good to say. Yesterday, I saw
26:52that the Financial Services Commission
26:54held an emergency meeting because
26:55circuit breakers were triggered on both
26:56major indices. This happened yesterday
26:58at 2:30 PM. Chairman Kim Byoung-hwan
27:00presided over a market inspection
27:02meeting. So that was what I was talking
27:03about yesterday. There was a time last
27:05July when people were asking if
27:06everyone had gone crazy. At that time,
27:08there was something I kept talking
27:09about. I said if they mentioned
27:11introducing the Stock Market
27:13Stabilization Fund, the market would
27:14immediately take off, but in the end,
27:16it didn't happen. Our country’s stock
27:19market stabilization fund has been
27:20established, but it has never actually
27:22been injected into the stock market.
27:25The structure is based on a "capital
27:26call" method; for example, we might say
27:28we’ll set up a 10 trillion won fund
27:30for emergencies, so we don't need it
27:32right now. But we secure agreements
27:35with financial institutions A, B, and C
27:36in advance, telling them we plan to
27:38raise 10 trillion won. So, when we say,
27:40"Hey, we think we need to act now, so
27:41can you provide what you agreed to?"
27:42Let's say it's 3 trillion won. "Bring 3
27:44trillion won each." If they deposit
27:45that, it immediately enters the stock
27:47market as a net buy, but they haven't
27:49made a capital call yet. They created
27:51the stabilization fund, but it has
27:52never actually been deployed. Why
27:53hasn't it been deployed? Whenever there
27:56was a sharp drop, the Financial
27:57Services Commission or the Ministry of
27:59Finance would just mention that they
28:00were considering deploying the fund,
28:02and the market would instantly soar, so
28:03they always solved it without lifting a
28:05finger. But in July, no such talk came
28:07out. If they were going to do something
28:09, even mentioning that they might do it
28:11—without actually having to deploy it
28:12—would cause the market to react
28:14immediately. That’s why we always saw
28:15V-shaped rebounds, and I said yesterday
28:17it would be good if something like that
28:18came up, but nothing did. In my view,
28:20the reason it didn't come out is one of
28:21two things. When I looked at what was
28:22discussed after the meeting, it was
28:23about refraining from reckless margin
28:24trading and debt-financed investing.
28:26Looking at this, I thought their
28:27understanding of finance was absolutely
28:29zero. I mean, if that’s all they were
28:31going to talk about, I can’t
28:33understand why they even held the
28:34meeting; this is something that should
28:36be handled by people with long
28:37experience in the industry and deep
28:38expertise in capital markets. I was so
28:41dumbfounded watching this that I
28:42thought it would have been better not
28:44to hold the meeting at all. Of course,
28:46I can understand it a bit from their
28:48perspective, as it was a meeting of the
28:49Financial Services Commission, the
28:51Financial Supervisory Service, and
28:52brokerage executives, so it might have
28:54been tricky for them to decide on the
28:56stabilization fund. Because to do
28:58something of that scale, you need
29:00someone equivalent to the U.S. Treasury
29:01Department. It should really involve
29:03our Ministry of Economy and Finance at
29:05the ministerial level, but if that’s
29:06all they were going to talk about,
29:08there was no need to hold it. They just
29:09went for a casual gathering. This
29:11simply needs to be criticized. It’s
29:13truly wrong, so although there were
29:14expectations, nothing resulted from it.
29:17And HLB Chairman Jin Yang-gon said he
29:19would decide on the FDA approval status
29:21as early as within the year. But we
29:23don't know how that will turn out.
29:26There are Class 1 and Class 2
29:27classifications; when you apply for FDA
29:29clinical approval, they categorize it
29:31as either Class 1 or 2 to decide
29:33whether to approve it quickly or take a
29:35long time to review. If it is
29:37classified as Class 1, theoretically,
29:39approval is granted within one to two
29:41months. So, it was said that it would
29:43be out within the year if classified as
29:44Class 1, but last time they said it
29:46would be Class 1 and it came out as
29:47Class 2, so we don't know yet. Please
29:49just use this for reference. Also,
29:50regarding the Halozyme multi-drug
29:52patent PGR currently under review by
29:53the U.S. Patent Trial and Appeal Board,
29:55two additional patents have been
29:56canceled. To put it simply, there are
29:5814 patent disputes between Alteogen and
30:00the U.S. company Halozyme currently
30:02before the Patent Trial and Appeal
30:03Board, and this means Alteogen has won
30:0511 of them. Then there are three
30:06remaining. The probability of winning
30:08these as well is high. The conclusion
30:09hasn't been reached yet, but it's
30:10highly likely they will win. And today,
30:12you should note that a buy-side sidecar
30:15was triggered on the KOSDAQ. So, that
30:17pretty much covers today's discussion.
30:19Yesterday, all sectors were simply poor
30:21, but today most were strong. Well,
30:23it's not all of them, but because most
30:25were strong, they rose evenly, and of
30:27course, some of them rose more strongly
30:28. These are representative examples,
30:31and what didn't rise today was
30:32secondary batteries, but it doesn't
30:34seem like there was any specific
30:35individual bad news for them.
30:36Everything else was mostly good. And
30:38looking at Samsung Electronics, when I
30:40started the broadcast earlier, I
30:42mentioned that the P/E of electronics
30:43stocks is three times, but even so,
30:45this just doesn't make sense. Looking
30:47back at July 29th here, that was
30:49exactly the day I was criticizing the
30:51KOSPI on air, asking how such a
30:52situation could even happen. I told you
30:55to remember this day, and the KOSPI
30:56ended up being stuck to it. That's
30:58because electronics stocks keep going
31:00down, but at this level, considering
31:02the valuation, a double bottom or a
31:04V-shaped rebound is normal. And
31:06fortunately, the KOSDAQ fell less; if
31:08it had fallen like the KOSPI, the
31:10KOSDAQ would have already broken
31:12through 700. It should be here, but
31:14it's not. This is because there were
31:16many days in the past when electronics
31:18stocks didn't rise, but materials,
31:19parts, and equipment stocks did. Or
31:21secondary batteries might have pushed
31:23it up on their own, so the level is
31:24still higher, but in terms of momentum,
31:26it would be right to focus on that side
31:27. So, for electronics stocks, at this
31:31point, it's actually just absurd
31:33because the P/E is only about 3.1 or
31:353.3. It's just too cheap. Also, when I
31:38was watching the market close earlier,
31:40I mentioned that one stock seemed to be
31:42the protagonist, and that was Celltrion
31:44. When our market crashed yesterday and
31:47triggered a circuit breaker, the ones I
31:49wondered about for not falling were
31:51Celltrion and Samsung Biologics.
31:53Samsung Biologics was up 5%, and
31:55Celltrion was up about 1-2%intraday,
31:57closing near flat at 0.5%; but being
31:59flat yesterday was a miracle. Since
32:02fewer than five out of 100 stocks
32:03stayed flat, they held up exceptionally
32:05well. So, were there any individual
32:07positive factors for them? There were
32:08none. There were no individual factors,
32:10so if I had to name one, it would
32:11obviously be the Treasury yield, which
32:12played a major role in the US market's
32:13rise. Since it has been diving for two
32:15consecutive days, it is only right to
32:16acknowledge it at this point. Also,
32:18there is an ETF that tracks the 20-year
32:21US Treasury yield exactly. That is TLT.
32:25Investors in US stocks will know this,
32:27but if you look at TLT, it was listed
32:29back in 2004. We are currently at an
32:31all-time low for the past 22 years.
32:33What does that mean? To put it simply,
32:35because its underlying asset is 20-year
32:37long-term bonds, it means bond prices
32:38have broken through all the lows of the
32:40last 20 years. For your reference, the
32:43previous time interest rates spiked
32:45significantly was October 2023. It is
32:47right here, exactly. At that time,
32:49rates spiked near 5%then started diving
32:52, and as bond prices rose and rates
32:54fell, the US market saw a three-month
32:56rally. Coins went up fivefold. During
32:58that period. So, we are looking at
33:00whether that same phase is emerging
33:02here now. There was only one thing that
33:04had a major impact on the sharp drop in
33:06bond yields. The US Treasury Department
33:08still hasn't intervened. Why was that?
33:11"Hey, even so, isn't a 10-year yield
33:13exceeding 5%offering too much interest?
33:15" To put it the other way around, bond
33:17yields rising this much means bond
33:19prices are too cheap. "Even if rates
33:22rise further, a 10-year yield at 5.5%is
33:24just too cheap." Investors think these
33:26bonds are a bargain, so bottom-fishing
33:28is coming in. It was a bit different
33:30back then. Back then, the Treasury
33:32intervened directly and twisted the
33:34long-term and short-term bonds, which
33:36is why it happened, but the structure
33:38is different this time. In any case,
33:40since it has rebounded strongly due to
33:42bottom-fishing, that would be a very
33:44good picture. So, looking at this,
33:47rather than just buying Celltrion/bio
33:49stocks, they always act like they're
33:51about to make a big move but then just
33:52die out, teasing us like that; and if
33:54they finally do rally, by the time you
33:56wonder why they're going up, other
33:58stocks will have already taken off. So,
34:01keep that in mind as a benchmark for
34:03next week. And even though everything
34:06was decent today, some of the ones that
34:08surged even harder are the ones listed
34:09here. They didn't have any specific
34:11individual news, either. For ECM skin
34:13boosters, Hans Biomed and C&G are the
34:14top two, and they both nearly hit the
34:16daily limit today. So they received a
34:19lot of attention, and although they
34:20didn't rise much today, it's been about
34:22one or two weeks since the market
34:24plunge. In that case, the stocks that
34:26were leaders right before the drop are
34:28highly likely to become leaders again.
34:30The representative theme for that is
34:32medical AI. So that would be stocks
34:34like Lunit, VUNO, 3billion, Protec, or
34:36See-ers. The market index started to
34:39plunge around here, but the trend for
34:41these stocks had already fully broken
34:43out and risen before that. So if you
34:46own these, you should obviously hold
34:47them, and if you don't, it's right to
34:49buy them. I think that's just about it
34:51for my comments. There’s really
34:53nothing else worth looking at. Because
34:56the market fell together yesterday due
34:58to the index and tried to rebound
35:00together today, I think this pretty
35:01much concludes the briefing. And I've
35:04also set it up so that content gets
35:07uploaded throughout the weekend.
35:10That’s because even when we’re
35:11resting, the U.S. market or the crypto
35:13market might move, or important
35:14financial news might break, so please
35:16do come in and check it out. So, watch
35:19the briefing before the market opens if
35:21you have time, and I’ve even set it
35:22up to provide feature stocks during the
35:24day and after the close. I'm still in
35:27the process of building this, so the
35:29testing isn't quite finished, but it
35:30should all be up and running starting
35:32next week. For example, by 4 p.m., it
35:35will show the day's featured stocks in
35:38the domestic market, which ones led,
35:40and why they rose; since my full-time
35:42job is stocks, I can watch all day, but
35:45I know many of you can't. So I upload
35:48market analysis in between like this so
35:50you can see all the key points of the
35:51day. For your reference, based on EPS
35:54forecasts, the year is already coming
35:56to an end. So, what’s the number one
35:58sector for net income growth rate next
36:00year? Did you know it’s secondary
36:01batteries? So I’ve organized that
36:03information separately along with the
36:04related stocks and uploaded it as well.
36:06I will keep uploading in that format,
36:08so it will become an all-in-one style
36:10resource where you can see everything
36:12from the U.S. and domestic markets to
36:14notable stocks during the trading
36:15session, making the content much richer
36:17. You can refer to that, and you've all
36:20worked very hard. I will see you next
36:22week. Have a great weekend. Thank you
36:25again for today. You did a great job.

About this transcript

This page contains the full transcript of [주식] 261016 이제 이 종목은 보고있어야 합니다 by 한균수의 주식사용설명서, generated from the public captions YouTube serves with the video. The transcript has 7,460 words across 1,124 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.