YT YouTube2Text

YouTube2Text → Transcripts → AI 거품 논란에 가려진 돈의 대이동, 지금까지의 투자 상식이 달라지는 이유 (이효석 대표 1부)

AI 거품 논란에 가려진 돈의 대이동, 지금까지의 투자 상식이 달라지는 이유 (이효석 대표 1부) — Transcript

by 채국장의 코스피2만행 머니버스 · 4,021 words · 690 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00People keep talking about China
0:01releasing things cheaply, but there are
0:03different levels of intelligence.
0:06With the exception of SpaceX’s XAI,
0:08OpenAI, or Entropy, virtually every
0:10other company is playing on the
0:12defensive.
0:14>> So, regarding the market in 2027, do you
0:16have a view on which sectors or
0:18industries will become the leading
0:19stocks? You must be
0:21>> overwhelmed keeping track of stocks,
0:21real estate, and ETFs,
0:25right? For such investors, there is the
0:28all-in-one investment advisory
0:29membership, Choi Sang-wook’s Money Bus
0:312.0.
0:33Share investment information and diverse
0:34opinions in a members-only real-time
0:36group chat.
0:38Solve everything you need for stocks,
0:40real estate, and ETFs all in one place
0:42with Money Bus, including monthly
0:44reports and special materials, live Q&A
0:46sessions where you can ask questions
0:48directly, and invited live sessions with
0:51current experts.
0:53Please refer to the link in the video
0:55description box.
0:57Hello. This is the interview with
0:58Director Choi Wook. The prediction that
1:00the era of AI agents is approaching as
1:02the AI industry develops rapidly is
1:04correct. However, the important question
1:06is, how will this change affect the
1:08stock market? So today, we are hosting
1:11CEO Lee Hyo-seok, who is the person best
1:13suited to cover this topic. Hello.
1:15>> Yes, hello.
1:17>> First of all, I would like to express my
1:18sincere gratitude that our shared
1:19passion has finally led us this far, and
1:20that you are appearing on our channel. I
1:22am not saying I am doing anything
1:23extraordinary, but I am truly thankful.
1:24Also, you published a book, didn't you?
1:26*Lee Hyo-seok's 21st Century Investment
1:27Method*. I even got an autograph on it,
1:29and the book turned out quite thick.
1:30>> Ah, I organized the content I have been
1:33diligently honing for about five years.
1:36>> Oh, and the photo turned out really well
1:37, too.
1:38>> Yes, yes.
1:39>> And looking at it, you have always had a
1:42strong interest in AI, haven't you?
1:44>> Right. Yes.
1:45>> And although there may be many people
1:47who don't know this, back in 2022 when
1:48we worked at the same company, you
1:50caught on to the issue very quickly when
1:52ChatGPT first came out. Starting in
1:54early 2023, even before Nvidia's stock
1:57price took off, you were actively
1:59promoting it. At that time, even those
2:02who thought, "Why all the fuss about AI
2:05?" were thinking...Around May or July of
2:072023, we started receiving a lot of
2:08feedback, with people saying thank you.
2:10That is
2:10>> correct.
2:12My interest in AI actually dates back to
2:14my graduate school days, specifically
2:17the year 2000. That is because the
2:19predecessor of AI is data mining.
2:22Back then, since there was no data, we
2:24referred to it as "mining" data. Now,
2:26with so much data available, we have
2:28entered the Big Data era, and since then
2:29, the AI era has begun.
2:30Since many of my friends work in AI or
2:32related fields, I think they naturally
2:34started hearing stories like, "Oh, is
2:36that really happening?"
2:38and I felt like I was talking about it
2:41in a realistic way, which likely sparked
2:43a lot of interest. That is
2:45>> correct.
2:46So, now that everyone has become an
2:48expert on AI, I believe you, Mr. Hyosung
2:50, possess a certain foresight to look
2:52ahead to the next era.
2:56Since we are currently at the end of
2:582026 and looking ahead to 2027, I think
3:00it is important to view the long-term
3:02picture today, even though short-term
3:04issues are important. If
3:06>> you ask a good question, I will give it
3:07serious thought. Please start with a
3:09>> good answer. Please
3:11>> ask a good question.
3:11>> Okay. I will ask a good question right
3:14away. First, regarding the results of
3:16the US midterm elections,
3:19one of the major issues was data centers
3:21. The market is being affected by delays
3:23in the pace of expansion or tightening
3:25regulations related to data centers
3:26within the US.
3:29However, while this cannot be seen as a
3:30factor damaging the long-term picture of
3:32AI, it is likely viewed as just a bit of
3:34short-term noise.
3:36Is that correct?
3:40I have
3:40>> thought a lot about how best to
3:42understand "data centers," and I think
3:45it would be appropriate to compare them
3:47to industries we are already familiar
3:50with.
3:51Furthermore, since the 20th century was
3:53the age of oil and the 21st century is
3:55the age of data computing, we are
3:57comparing them
3:58>> now, and the master Esmeric is coming up
4:00. Good. Yes. Let's try it
4:02>> this way.
4:03>> The 20th century was the age of oil. So
4:06, people dig for oil, right? In Saudi
4:08Arabia? They extract the oil. Then, they
4:10bring that oil to our country. Then, oil
4:13refineries boil it. We call that
4:16refining, right?
4:17Then, gasoline comes out, diesel,
4:19aviation fuel, asphalt, and various
4:21other products.
4:24Then, we use those to drive cars or
4:26airplanes.
4:28But if we apply this directly to that,
4:31if we think about where AI data centers
4:33are located, computing—the so-called
4:36intangible asset—is the oil.
4:39So, oil is computing.
4:41>> Yes.
4:42Ah, so crude oil?
4:45>> Unrefined crude oil is computing. Then,
4:47what is a data center?
4:49It is a data center that produces crude
4:52oil. Data centers produce computing.
4:55So, essentially, a data center can be
4:58compared to a place where crude oil is
5:00extracted.
5:02How did Saudi Arabia become so wealthy?
5:05Isn't it simply because people use the
5:07crude oil? Then, how much will computing
5:09be used in the future?
5:11If we consider this, isn't the reason
5:13the United States is building data
5:15centers by sucking up money from all
5:16over the world like a Spirit Bomb their
5:18intention to play the role of the Saudi
5:20Arabia of the 21st century?
5:23If we think about it that way, it isn't
5:25the Democratic or Republican party that
5:27matters; we need to think about the
5:28significance that data centers hold in
5:30the 21st century. Although the
5:32>> question was quite crude, the answer
5:35reeked of a master.
5:37A master of this caliber isn't just
5:39someone who worries about the world; he
5:43is a very visionary master.
5:47There are figures like Philippe LaFont,
5:48and you can sense the aura of a master
5:50of that caliber.
5:51To connect that further to data centers
5:54,
5:54>> data centers provide the "crude oil"
5:56known as computing. Now, there are
5:58places that take that and refine it for
6:00use, you know. There are oil refineries
6:02and chemical companies. If you look
6:04closely, almost all the products we see
6:07originate from oil. So, even plastic is
6:09produced through refinement, and there
6:11are so many different kinds, starting
6:12from naphtha.
6:14You probably know all of this very well
6:16, especially since you covered
6:17construction and how buildings are built
6:18.
6:19But if you think about which companies
6:21play that role, it is Entropic and
6:23OpenAI.
6:24XAI is the same.
6:26So, they use computing to refine and
6:28create intelligence.
6:30>> Yes.
6:31>> This intelligence is like gasoline,
6:33kerosene, or diesel.
6:34But just as there is a wide variety of
6:37high-quality gasoline, just any fuel
6:39that works, and asphalt, I think we need
6:42to realize that intelligence also has a
6:45hierarchy.
6:47So, people keep talking about China
6:49releasing things cheaply, but
6:50intelligence also has grades.
6:52Therefore, the grade that Entropic
6:54becomes is high-quality gasoline.
6:55>> Ah.
6:56>> So, like that asphalt thing, even
6:58asphalt has its uses. But that
7:00>> analogy was crazy.
7:02>> But if you look at it that way, the
7:04industry makes sense.
7:06And finally, where do you put this oil?
7:09You put it into a fuel tank, right?
7:10>> Yes.
7:11>> That is why we drive cars, isn't it?
7:15So, once you understand this picture,
7:17you can think of people using AI and
7:18Entropics refining oil to create
7:20intelligence, right?
7:24Then, last week on Orin's podcast, I saw
7:27someone explain Entropics 'S1, and it
7:29was a brilliant explanation.
7:33Let me explain it by connecting it to
7:35numbers.
7:36Now, let's say Nvidia takes a $ 1
7:39component. That $ 1 cost likely includes
7:43HBM and everything else.
7:46To supply that data center component,
7:48Nvidia takes that $ 1 and processes it
7:51to turn it into a $ 4 component.
7:54Because the GP margin is about 80%. It
7:57just roughly becomes $ 4. So, who buys
7:59the $ 4? It's Elon Musk of SpaceX. He
8:02buys it, brings it in, and builds data
8:04centers. But regarding what he makes by
8:06building data centers—I mentioned
8:09earlier that he makes computing—it seems
8:11he sells it for roughly $ 8 to $ 10. So
8:14, that means he's making a profit of
8:15about double the cost.
8:16>> So, is that in the data center? Or in
8:19the computing?
8:21>> Well, if you take an Nvidia chip, attach
8:23all sorts of components to it, and
8:24calculate it based on cost, doubling the
8:26production value, he sells it for about
8:28$ 8 to $ 10.
8:31And then, regarding Entropy, which is
8:33important here, how much is Entropy
8:34making a profit from taking the $ 8
8:36computing unit and turning it into a
8:38product to sell?
8:40This is very important.
8:42Looking at the data released at S1, they
8:45mentioned that roughly $ 1.8 to $ 2 in
8:48revenue is generated per $ 1 of
8:51computing unit.
8:53Here too, he's making a profit of double
8:54the cost.
8:56Then, regarding those who take
8:57intelligence, if they double their
8:59profit, that would be about $ 15, right
9:01?
9:02So, it’s $
9:03>> 15 or $ 20. Yes. The question is,
9:05>> how much value do they perceive in that
9:08?
9:09This is very important, and we really
9:11need to think deeply about it here. It
9:13is difficult to calculate. Because how
9:15do you calculate the satisfaction felt
9:17when driving a car? It’s like that.
9:20People fill up with gas, think, "Oh, I
9:21bought a car and it feels good." But
9:23what is the value of that?
9:24>> Utility, utility. Yes. We
9:26>> cannot know that. However, if we look at
9:28it now, it is the same for people who
9:30buy and use intelligence.
9:32They use intelligence to make songs,
9:33write stories, edit YouTube videos, and
9:35just enjoy it.
9:37But what is the value of that? It seems
9:39we still haven't been able to calculate
9:40that.
9:41But in my opinion, I think they are
9:43likely perceiving a value of at least
9:45two to ten times that amount.
9:49So, for example, imagine buying a car.
9:51If you think about how many kilometers
9:52it travels per liter of gasoline and how
9:53much utility you feel from putting in
9:55just a certain amount of money, you get
9:56the gist of it.
10:00But the reason I am saying this is that
10:02many companies in this so-called value
10:04chain are actually all making money
10:06right now.
10:09So, of course, the reason someone needs
10:11money is because there is a huge need
10:14for data centers and computing power.
10:17However, in this entire market, it isn't
10:20like someone is making massive
10:21sacrifices, generating huge revenue
10:24while suffering huge losses; rather, the
10:26players here are actually making money.
10:30I think that is very important.
10:33>> Ah, so the crude oil analogy really
10:35resonates with me.
10:37Even though it is currently around $ 100
10:39per barrel, the cost of crude oil itself
10:41is dirt cheap.
10:44So, just like with crude oil, there is
10:46the drilling margin, the refining margin
10:48for the initial refinement, and the
10:50product spread for making goods from
10:52that refined product.
10:54Similarly, in computing, if you look at
10:56the case of Nvidia, which buys GPU
10:58infrastructure, they profit from that
11:00margin based on a kind of drilling
11:01concept—and that is enormous, judging by
11:04the scale.
11:05But then there is the margin for data
11:07centers that build computing systems
11:08using that infrastructure.
11:09>> Hmm.
11:09>> This is the field where SpaceX is
11:10currently leading the way, and that
11:11margin is almost double.
11:15Then there is the computing margin for
11:15companies like Entropic or OpenAI that
11:16build computing systems using that
11:17infrastructure. Hmm. That is also double
11:19again.
11:21Ultimately, we are the consumers—
11:23companies or consumers—who use it.
11:27Of course, we are paying for it, but
11:29from the perspective of the payer, as
11:31you mentioned, we use it because there
11:32is utility in paying for it, whether it
11:34be cost savings or revenue growth.
11:36>> That is correct. Yes. It
11:38>> is very accurate to compare the value
11:39chain with the ratio of crude oil. Oh,
11:41when did you start thinking about this?
11:42I've actually been
11:44>> talking about this on my channel for
11:45about a month or two now, but after
11:46giving it some thought, it's actually in
11:47the book.
11:49It is in the book.
11:51But if you think about it carefully, the
11:53biggest concern is whether it will move
11:55forward, not move, or just crash.
11:58You're worried it might slip and fall
12:00while moving. And that could happen.
12:03Because even if you've built data
12:04centers, if the people using that
12:06intelligence don't drive the cars, that
12:07becomes a problem, right?
12:09That's correct. Not driving is a problem
12:11, and another factor is
12:12>> demand—final
12:12>> demand. That is
12:13>> final demand.
12:14Also, if you think about what was needed
12:17for cars to move around, you need
12:19highways.
12:20>> Highway tires.
12:21>> Yes.
12:22>> You need them to drive cars, don't you?
12:24>> Michelle used to tell you about good
12:26restaurants and act as a guide to make
12:27people travel, didn't she? Right.
12:30>> So, if you think about it carefully, the
12:32question is whether the company is
12:34paving the highway where AI can run
12:36around freely.
12:38But while we were laying the groundwork
12:39, the lack of a highway prevented AI
12:40from working freely. In fact, if it
12:42could just operate as it pleased, it
12:43would possess immense value. However,
12:45due to various constraints—such as
12:47security issues and other obstacles—it
12:49is currently experiencing a bumpy ride.
12:51While such processes are inevitable, I
12:54believe the industry has advanced so
12:56dramatically that there is no turning
12:58back; there is no going back. That is my
13:01view.
13:02>> Then, regarding what you mentioned
13:03earlier—whether we should view it as a
13:05value chain or not—that
13:06>> massive value chain is truly...yes. It
13:08is a
13:09>> kind of discussion about the hierarchy
13:11dominating the world.
13:13We have moved from the era of oil to the
13:15era of intelligence, and you discussed
13:17everything from intelligence
13:18infrastructure to demand.
13:20There are GPUs, data centers, computing
13:23, and the end-users, and market noise is
13:25emerging from every single one of these
13:28areas.
13:30It seems to be happening everywhere, but
13:32in your view, Mr. CEO, where do you
13:34consider the most fundamentally
13:36important aspect?
13:38>> Uh, but actually, while you mentioned
13:40that everything is a problem right now
13:42in the process of the AI revolution, if
13:44we think about how we got to where we
13:46are today, we couldn't have reached this
13:48point just because GPT was released at
13:50the beginning.
13:52The point is, what would have happened
13:54if Nvidia hadn't been releasing new
13:55chips like crazy every year?
13:57>> Yes.
13:58>> But if you think about it carefully,
14:00coming back to the crude oil analogy,
14:02how much better crude oil you can
14:04extract is technology.
14:06>> Yes.
14:07>> And how well you take that good oil,
14:09refine it with good technology to
14:10increase efficiency, and extract
14:12intelligence is also technology.
14:15And how you make cars with good fuel
14:17efficiency—that is also technology.
14:20Building highways, as I mentioned
14:21earlier—that is also technology.
14:24So, what is actually happening in this
14:26landscape right now is that all economic
14:29actors will start working like crazy to
14:32grow the industry, and to make it more
14:34efficient
14:36.
14:37>> Yes. Things will get cheaper and better
14:38and
14:39>> better.
14:41So, I believe that as better and better
14:42intelligence is developed more cheaply,
14:44and as we use it, an explosive
14:46revolution will occur, and I think we
14:47are right at the very beginning of that
14:49process.
14:50Yes.
14:51>> But since you used the analogy of cars
14:52and gasoline, it really resonates with
14:53me.
14:55In a way, among those who thought cars
14:57would become a means of personal
14:58transportation in the past, Ford, which
15:00created the S model, really took the
15:02lead. It was originally a very high-end
15:03product.
15:08In that context, while there are a few
15:10large corporations currently
15:12incorporating AI, if we call the use of
15:15it "AX," it doesn't seem like AX has
15:17progressed that much yet.
15:20It's like the
15:21>> highway hasn't been paved yet, so cars
15:23are just crawling along. The cars have
15:25been released, but they are crawling.
15:27The car's performance is excellent, but
15:29you know that red anatomical technique
15:31we use? It's like a person is standing
15:32in front, telling you to go slowly,
15:33saying you'll have an accident. That's
15:35how it is.
15:36Regarding what you said about Ford, they
15:38also initially made cars, but people
15:39didn't have enough money to buy them.
15:42And you can't even play enough to do
15:44something with your car. So they say, "
15:45Let's implement a five-day workweek."
15:47And then, "Oh? Let's pay a little more.
15:48" It’s like that. So, actually, the
15:50discussions we're having now—like basic
15:52income or AI—are much larger in scale
15:54when viewed within the broader framework
15:56.
15:57But I get the feeling that things are
15:58currently flowing in a similar context.
16:02>> Yes.
16:04While thinking about various things
16:05during the process of setting the tone,
16:07when considering what the biggest
16:09bottleneck might be next year, it seems
16:11that laying the infrastructure to spread
16:13AX widely is crucial.
16:15That's right. Ultimately, if the final
16:18fleet is
16:18>> strong, no one can refute it. And that
16:21strength has to be systemically strong,
16:23doesn't it?
16:24It shouldn't just be a one-off move; the
16:26entire system itself needs to be
16:27established.
16:29Then, strictly speaking, whether it is
16:31individuals, large corporations, or SMEs
16:33—in other words, whether it is B2C, B2B
16:35, or B 2 G—how actively will these
16:37customer segments use AX?
16:40What are the resistances to using it,
16:42and who is resolving those resistances?
16:44Also, since companies looking to make
16:46money by selling computing will be
16:48formulating and pushing their own
16:49massive strategies in these areas, it
16:51seems like 2027 might turn out to be a
16:53very rough year, especially if intense
16:55competition is expected to emerge. That
16:56is
16:57>> correct. And if I were to
17:00>> ask the question about 2027 right here,
17:01from your perspective as someone
17:03concerned with demand-side activation,
17:05infrastructure, and user-level
17:06considerations, do you have a view that
17:08which sector—whether it be the leading
17:10stock, the leading industry, or the
17:11leading company—will emerge in the 2027
17:13market?
17:15>> It is security, security.
17:16>> Ah.
17:16>> Yes. It is simply security. It is just
17:19so clear.
17:20We refer to this as "technical debt."
17:23Technical debt. To use an analogy,
17:25imagine attending multiple academies and
17:28taking many classes; if you don't review
17:30, nothing remains.
17:32It’s similar to that.
17:34Imagine technology from Entropy or
17:35OpenAI comes out with such great
17:37features that developers boast about how
17:39much they can do.
17:41However, when they try to apply it, they
17:43realize the security is too vulnerable.
17:46Since security hasn't kept pace, this
17:48vulnerability lags behind, which we call
17:50"alignment."
17:52So, there is a distinction between
17:53technology and usable technology—that is
17:55, using good technology versus the
17:56ability to use it.
17:58To give you an example since we keep
17:59talking about cars, if you ask what is
18:01needed for a car to run smoothly, there
18:02are actually a surprising number of
18:04trivial details.
18:05For instance, there's the rule against
18:07drunk driving.
18:08>> Yes.
18:08>> That wouldn't have existed originally,
18:09right?
18:10>> That driver's license.
18:10>> Things like a license are all necessary
18:13.
18:13So, in a way, new technology has emerged
18:15, but there is technical debt that must
18:18be overcome during the application
18:19process.
18:22Since resolving that technical debt
18:24mostly involves security-related matters
18:26, I believe that security is currently
18:28the biggest bottleneck in the transition
18:30to the application side.
18:33>> Ah. So, because the full-scale expansion
18:34of AX is actually being hampered by
18:36security, security is the true
18:37bottleneck; once that is resolved, many
18:39issues will be resolved based on that
18:41foundation.
18:42>> Right, quite a lot of issues will start
18:44to be resolved.
18:45So, I believe security is what is needed
18:47for those issues to start to materialize
18:49.
18:49>> Yes. Then, conversely, when we enter a
18:51computing era where massive user-based
18:54AI usage increases, what will be the
18:56ones that die? For
18:58>> example, regarding user-based companies
18:59, there was
18:59>> talk about whether software firms would
19:00die or not, then people said that didn't
19:01seem to be the case, claiming they
19:02wouldn't die because customers are still
19:03strong.
19:07But then stock prices started
19:08fluctuating wildly again, so there seems
19:10to be a lot of optimistic talk about
19:11this. In this regard, what exactly will
19:13be the ones that die?
19:16>> Rather than saying that a specific part
19:18dies, I would prefer to say that if you
19:20do nothing, you just die.
19:22Right now, it’s not about what you’ve
19:24done in the past—or whether you’re the
19:25type of person who does what—but rather
19:27, if a company just sits still, it
19:28simply dies. If you
19:30>> don’t do AX, you die?
19:31>> Yes. So, rather than defining things by
19:33saying, "Hey, you did software, so what
19:35?", it seems that if you don’t change
19:37quickly, everything just dies. That’s
19:39the thought I’ve had. So, just because
19:41it’s a software company doesn’t mean
19:43every one of them will survive.
19:44However, whether there is a "value"
19:46within that context might depend on this
19:48: Do I really need to build that for AI
19:50?
19:51Well, you can build it, but if you think
19:52, "I can just use it without building it
19:54," then you might survive. But I don't
19:57think that itself will be a significant
19:59"value." Because it involves going
20:01through an intermediate step, the
20:03software becomes something that AI uses
20:05, so the AI itself is more valuable, but
20:07the software that AI uses might not
20:09necessarily be valuable. They won't die
20:11, though. But that doesn't seem to have
20:12any investment appeal.
20:14Instead, regardless of the management or
20:16the organizational culture, isn't it the
20:19case that organizations that change
20:20quickly survive, while those that don't
20:22die?
20:23That's why I heard that a very important
20:27keyword in Silicon Valley in 2027 is "
20:31reform." Repage? It seems they keep
20:34saying that the whole thing needs to be
20:35changed.
20:37So, that is a huge issue at Google as
20:38well; even internally, Google needs to
20:39change everything right now, doesn't it
20:41?
20:42>> Yes. Isn't that the case?
20:43>> The reason they need to
20:44>> change everything is that Google's core
20:46business used to be search-based, with
20:48ads attached to search results. But
20:50recently, if you look at it, far more
20:53people are searching on LMM. These days
20:55, I also ask everything on Chat GPT or
20:57Co-pilot.
20:58>> That's right. People
20:58>> don't bother opening Chrome, and if that
21:00side gets taken over, the core business
21:02will falter. So, as you mentioned
21:05earlier—reform, repage—the entire
21:07company just needs to change.
21:09>> The question is whether the main base
21:11can defend itself right now.
21:14In fact, if we consider which companies
21:17are attacking with AI, I think that,
21:20with the exception of SpaceX’s XAI,
21:22OpenAI, or Entropy, virtually every
21:25other company is on the defensive.
21:29>> Ah, yes,
21:30>> defense. Because even Google is
21:32currently under attack due to AI and is
21:34trying to defend it even more. And the
21:37same goes for Amazon. Amazon told Weizhu
21:39, "Don't come onto our territory." So,
21:41they are all defenders. From my
21:43perspective, it seems the only attackers
21:46are Entropy, OpenAI, SpaceX, and Meta.
21:51>> Ah, so next year will be a year of
21:53fierce battles between offense and
21:54defense, right? It
21:56>> looks like there will be a massive
21:57battle.
21:58>> Should investors just stand by and watch
21:59? What should they do?

About this transcript

This page contains the full transcript of AI 거품 논란에 가려진 돈의 대이동, 지금까지의 투자 상식이 달라지는 이유 (이효석 대표 1부) by 채국장의 코스피2만행 머니버스, generated from the public captions YouTube serves with the video. The transcript has 4,021 words across 690 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 choose a caption track available on the video.

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, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.