YT YouTube2Text

YouTube2Text → Transcripts → Der KI-Hype hat die Realität überholt!

Der KI-Hype hat die Realität überholt! — Transcript

by Balthasar Becker · 4,720 words · 718 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:01Anthropic and OpenAI will perish
0:03because of the revolution they ignited
0:05themselves. We now have another three
0:09to six months of massive price gains
0:11ahead of us before the AI trade
0:14implodes. And with that, welcome to the
0:18great AI trap. In today's video, we are
0:22looking at why OpenAI and Anthropic
0:25might be financing the revolution, but
0:28will perish because of it. Which stocks
0:32will continue to benefit most from the
0:35AI trade. We are going to witness a
0:38transformation of the AI trade. Moving
0:41away from the bottlenecks toward the
0:43application, where AI actually finds a
0:46use. And we will look at the past boom
0:50and bust cycles of telecommunications,
0:53the railroad, and fiber optic expansion
0:55to learn how such a technological
0:58revolution happens. Because it always
1:01happens according to the same pattern.
1:05The stocks, the companies that finance
1:08the whole thing, are usually not the
1:10companies that end up making the money.
1:14And with that, we start with the
1:16uncomfortable AI calculation. We had
1:19the Capex earnings and we see that 16.4
1:22billion dollars are being invested.
1:26Operationally, however, the AI business
1:29only made 430 million dollars, or 0.43
1:35million, 430 or 0.43 billion dollars in
1:38profit. That means the factor is 40 to
1:421. And that is very reminiscent of old
1:47patterns, just like the fiber optic
1:49boom. And we all know what happened.
1:53The Nasdaq dropped by 77%from March
1:562000 to the fall of 2002. It collapsed.
2:02And that is a price drop that nobody,
2:05absolutely nobody among us can even
2:08imagine, that an index like the Nasdaq
2:12could fall massively again in a
2:14multi-year bear market, no one can
2:17imagine that. And back then, the
2:20technology was right. Very, very right,
2:24very, very clear. The internet was the
2:26future. Everyone was aware that the
2:29internet would be the future.
2:31Everything would be. Only the investors
2:35who paid for the construction, who laid
2:37the fiber optics in the ground, were
2:39not the ones who could monetize the
2:42whole thing. And we are going to look
2:45at that now. What you will learn in
2:48this video is where we currently stand
2:50in the hype cycle. Then, which factor
2:55is currently deciding whether the hype
2:57will break in the next three or the
2:59next 18 months, because it will break
3:02at some point; every hype will break.
3:05The only question is, are we currently
3:06at the beginning, in the middle, or at
3:08the end? Investing is always about
3:11riding the trend. But anyone who surfs
3:13knows that at some point, the wave
3:15flattens out. And when the wave has
3:18flattened out, you don't want to be the
3:21last one invested in these AI stocks.
3:25And we derive that by learning from the
3:28cycles of the past, and then we will
3:30determine where the money flows next.
3:34And that is exactly what we want to do
3:36here, to achieve excess returns. So
3:38folks, the value chain—you have to
3:40understand how this AI topic works.
3:43It’s insane how much misinformation
3:45is being spread here, from "it's all a
3:47bubble." To "the world is ending
3:50tomorrow." We will all succumb to AI,
3:54all the way to a lot of drama.
3:57Ultimately, the deciding factor is
4:00understanding the value chain, where
4:03the bottlenecks are, where the money is
4:05earned, but also who is funding the
4:08whole thing, because that is a huge
4:11problem. We essentially have four
4:14building blocks that we are going to
4:16take a quick look at now. It all starts
4:19with the AI models. Anthropic, OpenAI,
4:23Google Gemini, Groq. These are the AI
4:26models. Very, very important. Then you
4:29have applications, i.e., software
4:31companies that, for example, integrate
4:33the models into their products. For
4:37example, a software company that,
4:40because it uses AI, asks for more, but
4:43hopefully also improves its margins and
4:46productivity, or you use AI—I, for
4:49example, have nicer slides thanks to AI
4:52. My slides used to be ugly; now with
4:55AI, they are suddenly beautiful. Then
4:58we have the data centers; they are very
5:00, very important. These are, um,
5:03actually the places where all the power
5:06—the compute—that which ultimately
5:09makes AI possible, provides that
5:11computing power. And we have the
5:15infrastructure manufacturers who build
5:18the whole thing, and there is a flawed
5:20financing loop here, and we're looking
5:23at that now, because ultimately the
5:25whole thing works like this: we have
5:27the AI labs. For example, Anthropic,
5:31OpenAI, who say: "With AI, we will
5:34change the world." Then they go to the
5:36investors, and both are not yet on the
5:38stock market. They go to private
5:40investors and say: "Hey, with AI we are
5:43changing the world, give us money." We
5:45have to invest a lot of money now in
5:48order to benefit from this new
5:51technology. Then the private investors
5:54—they are not yet listed on the stock
5:56market—go to OpenAI and Anthropic and
5:58give them billions, a whole lot of
6:00money. This means the investors provide
6:03the capital. The AI labs demand the
6:07computing power, specifically from the
6:10data centers. And the data centers, who
6:14do they belong to? They belong to, for
6:16example, Amazon, they belong to Google.
6:18Very important. That means there are
6:20the companies that produce the data
6:22centers. Not produce, but financed and
6:26built them, and they have the AI labs
6:28as customers. And funnily enough,
6:33Amazon and Google are also invested in
6:36these AI labs. That means they gave the
6:41AI labs money, and with this money, the
6:43AI labs then, so to speak, secured
6:46commitments from operators like Amazon
6:48and Google to ensure that—and this
6:51will be important later—a lot of this
6:54AI computing power will be purchased
6:56from those data centers in the future.
7:01There are long-term contracts, which is
7:03why demand is so high and new data
7:06centers have to be built, and the
7:08suppliers are obviously benefiting
7:10massively; because the suppliers
7:12currently have much higher demand than
7:14they can provide in supply, margins and
7:17profits are exploding, and the stock
7:19market is soaring. The big risk in this
7:24whole cycle is that the AI labs are
7:27currently not yet making any money.
7:30They are losing money, and we will take
7:32a look at the numbers in a moment. They
7:34are not yet truly profitable. That
7:37means the cost of building this AI
7:39infrastructure is currently still
7:41higher than the money being earned with
7:44it. And that is the huge problem,
7:47because a moment can come at any time
7:49when the investors of OpenAI and
7:52Anthropic say, we won't give you any
7:54more money. And because we aren't
7:56giving you any more money, they stop
7:58requesting computing power because they
7:59don't have any funds left. And then
8:01this whole cycle collapses. Then there
8:05is way too much computing power, a lot
8:07of data centers that were built but are
8:09no longer in demand, no new data
8:11centers are being built, and all the
8:13suppliers that have been driving the
8:15stock market rally so massively will
8:16all crash. That is the important topic.
8:20So, everything here is currently
8:21hanging on two factors. One is whether
8:23they are building, or actually
8:26everything depends on one factor. Is
8:29the demand for computing power from
8:31Anthropic and OpenAI continuing to rise
8:33? And if it continues to rise, will new
8:38data centers be built? Then the AI
8:41trade will continue exactly as is. And
8:43if demand collapses because, for
8:45example, they aren't making money and
8:47no one wants to give them funds anymore
8:50, then the whole thing falls apart.
8:54Anthropic wants to go public in
8:56November, they want to raise something
8:58like 130 billion, and it continues
9:00later on. Yes, and we have massive
9:04revenues. We have 50 billion here. We
9:08can see here how the revenue growth
9:11increased each quarter for Anthropic:
9:15Q1 4.7, Q2 10, 11.6, Q3 16.8. That
9:19means the growth is there, but
9:21profitability is not really there yet.
9:24And the big question is, who is
9:25financing all of this growth? And that
9:28is exactly what we just discussed.
9:31Growth is being funded by investors in
9:33these companies, but the profits just
9:35aren't there yet. We do see a positive
9:40trend, we see that the share of
9:43Anthropic's revenue spent on computing
9:46power is falling. That is obviously
9:49good. That was still 144%in Q2 2024.
9:53Now it's only 50%. That's a good start.
9:57Then, theoretically, they would have an
9:59operating profit in Q2 2026. Quite
10:02honestly, um, you can look into the
10:04numbers, and there is an incredible
10:06amount of trickery going on. That means
10:09the direction is okay, but it is not
10:11the case that massive profits are being
10:14made right now. Above all, Anthropic
10:17has entered into 570 billion in compute
10:20obligations, and it will be interesting
10:23to see who wins—namely, the one
10:25selling that computing power to
10:27Anthropic. And that is essentially
10:29Google and Amazon, and primarily Google
10:31and Amazon also hold the shares that
10:33account for a total of nearly 26%of
10:35Anthropic. That means when Anthropic
10:38goes public, they will be the ones
10:39profiting the most again. Um, here we
10:42see Anthropic earning money
10:44operationally, but I don't want to dive
10:47too deep into the whole balance sheet
10:49topic here. The essential factor is
10:52that common—but, um, yeah, then the
11:01big question is, what happens when no
11:03new money comes in? So, if, for example
11:07—and that's why you have to look at
11:08the earnings from Amazon, um, at these
11:10earnings reports from Google, Amazon,
11:12and so on—will they continue to spend
11:14more money? SpaceX said they will
11:16continue to spend more money. Tesla
11:18said less, Meta wants to continue to
11:20spend more money. That means it looks
11:22like the AI trade is currently just
11:24continuing to run. But the crucial
11:26question is whether the big companies
11:29like Amazon, Microsoft, and Google will
11:31keep spending more money on data
11:33centers or if they will eventually say
11:36they've spent enough. And then another
11:39factor that is very, very important is
11:41whether these models will become
11:43profitable. A technology can be
11:46extremely successful in the long run
11:49while its funding collapses at a
11:51certain phase. And that was exactly the
11:55case with the internet, where the
11:57funding collapsed, because higher
11:59revenue, um, is not everything that
12:01counts. We see it with SpaceX; they are
12:05investing 16.4 billion. Theoretically,
12:09they also made a profit of, um, 2.7
12:12billion after depreciation. You just
12:15have to say, the AI division only made
12:18those 420 million. The 2 billion and
12:21something were Starlink. So, uh, here
12:25you can see that for SpaceX, the whole
12:28thing is still ongoing, but you can see
12:30right here that, um, the profit is
12:33coming from Starlink so far. And the
12:37big question is simply whether the
12:39revenue covers the obligations, whether
12:41enough money can be earned, whether AI
12:44can become profitable enough to justify
12:46these gigantic investments. Usually,
12:49it’s the case that you have a new
12:51technology. New technology means an
12:54incredible need for investment and no
12:56profits yet. Now all this is being
12:59invested, and subsequently, the
13:01question is who makes the profits, and
13:02my thesis there is that the probability
13:04is very high that OpenAI and Anthropic
13:06won't be the ones making the profits.
13:09They financed the whole thing, and
13:11we’ll look into historical parallels
13:13in a moment, but that doesn't mean they
13:15benefit from it the most. We see here,
13:18Anthropic has obligations for 570
13:23billion, 67 billion dollars over the
13:25next few years to purchase computing
13:27power. And they are doing that
13:30primarily with SpaceX, Google, and
13:32Amazon. That means those are the
13:34companies making the money. Funnily
13:37enough, Google and Amazon, in
13:39particular, also invested early in
13:40SpaceX. That means they hold about 26%
13:43of SpaceX combined. If SpaceX goes
13:45public now at an inflated valuation of,
13:48say, 2 trillion, Google and Amazon will
13:51naturally make a lot of money if SpaceX
13:54rises. And that is exactly this nested
13:56financing that can also become a
13:59problem at times, but, um, I currently
14:01see SpaceX, Google, and Amazon as the
14:03winners, not OpenAI and Anthropic. And
14:07if you want to know exactly where we
14:09stand, how I’m implementing this in
14:11my portfolio, and how I’m betting on
14:13the end-of-year rally, I’m holding a
14:16webinar on the 12th. November at 7:00
14:19PM. It is 100%free. The link is in the
14:22video description, more on that later,
14:24but let’s move on. Every technology
14:26always goes through the same curve, and
14:29that is the Gartner Hype Cycle. It
14:32begins with a trigger; that was ChatGPT
14:34. Then expectations are driven upward,
14:40and at some point, reality catches up
14:42with those expectations. That means the
14:45technology eventually can no longer
14:46deliver. Then comes the trough of
14:47disillusionment. In this trough of
14:49disillusionment, the real progress
14:51happens. That’s when the profits are
14:53made, and then it continues. And that
14:55was exactly the same during the 2000
14:57dot-com bubble. The internet drove
14:58expectations. Up here, the internet
15:01couldn't meet those expectations. No
15:03money was being made. At the same time
15:05—and this is the deciding factor—
15:07the hype is always ended by a
15:08tightening of financing conditions.
15:10More on that later. Back then, the oil
15:12price rose along with interest rates,
15:14and that strangled the whole thing.
15:15There was no more money left, and then
15:18came this massive correction. During
15:21the correction, the profits started
15:23coming in, and um, things went way up
15:25for companies like Emerson. But what
15:28most investors won't be able to handle
15:30emotionally is this valley of tears.
15:33And let's take a look at where we are
15:35right now. So, we can see here that AI
15:39hasn't really arrived yet, uh, in a big
15:42way. We see that 89%of executives don't
15:46see any measurable productivity effect
15:48yet. As I said, there are various
15:51statistics. There are also statistics
15:53that say we are already seeing
15:54productivity gains from AI. Um, so it's
15:57not entirely clear yet. Currently, we
16:00are in the first phase, the phase where
16:02investment is happening because the
16:04hype is there. The insane productivity
16:07gains are the second phase. That will
16:09be the phase that is massively
16:11interesting on the stock market, where
16:13massive economic growth emerges and
16:14incredible profits and earnings growth
16:16for stocks will occur. That will be the
16:19best phase on the stock market, but we
16:21just aren't there yet. And we see that
16:26about half of the S&P are AI stocks,
16:30and if the bubble bursts, the S&P will
16:34fall massively. And that is very
16:36important to understand. And now let's
16:38take a look at other technology cycles
16:40and what we can learn from them. We see
16:43, for example, in the dot-com bubble,
16:46that a massive amount of fiber optics
16:48was laid, but investments collapsed,
16:54even though there was plenty of, uh,
16:57fiber optics. Investments collapsed,
17:01and those companies that produced it,
17:04that laid the fiber optics, like
17:06Telecom and so on, they crashed
17:08massively. The winners were those
17:12companies that could then, uh, benefit
17:15from it, like Google, Amazon, Netflix,
17:19and so on, which used the fiber optics,
17:22but not the ones that laid it. They
17:25were hyped. That was the bottleneck in
17:27the investment phase. Just like with AI
17:30right now. All the companies that are
17:32being hyped right now will presumably
17:34no longer be hyped once it moves into
17:36application. Um, how big is the AI boom
17:39really? People usually compare
17:42investments relative to GDP. Then you
17:45see, railroads were 3%of GDP. Currently
17:48, for AI, it's about 1.5%, if you take
17:51the Q2, uh, Q2 level. The telecom boom
17:55in 2000 was 1.4. Uh, electricity supply
17:59in 1920 was 0.9%. Um, if you take the
18:04investment plan for the entire year
18:062026, we are already at 2.3%. This
18:10means, uh, we are no longer at the
18:12beginning. And what you also see, and
18:17this is very, very important, is that
18:20the price for AI for tokens is falling
18:23massively. And when the price for
18:26tokens falls, that is initially good
18:29for those who want to use tokens, but
18:32bad for those who sell tokens. OpenAI
18:35and Anthropic sell tokens. For them, it
18:38is bad when the price falls. For those
18:40who want to use tokens now, it is good.
18:43And the token price is ultimately the
18:46price for AI, uh, for intelligence. And
18:49when the price for intelligence falls,
18:51more intelligence will be available.
18:53That means there will be incredible
18:55applications taking place. The more
18:57intelligence available in the world,
18:58the more extreme the applications will
19:00become. And the fact that the price is
19:04falling is good, just not for OpenAI
19:07and Anthropic, because they have to
19:10compensate for their demand by a higher
19:14factor than the price drop. If the
19:18price falls by 80%, they must have five
19:20times the demand for their tokens to
19:22have the same revenue. And that is
19:24exactly the big question where I say,
19:26with OpenAI and Anthropic, I see the
19:29big, big risk. Additionally, there is
19:31of course the problem that the hardware
19:33ages relatively quickly. That means you
19:35have very high depreciation, as we also
19:37saw with SpaceX. 50%of the profit was
19:40depreciation, and you have high running
19:43costs. This means that in the first
19:48phase, stocks are hyped that produce
19:50the whole infrastructure and stocks
19:53that sell AI like OpenAI and Anthropic,
19:56but that does not mean these will be
19:59the stocks that benefit in the long
20:01term. And now we are getting into the
20:05topics that could end this boom. We are
20:07seeing that capital is becoming more
20:09expensive. Government bond yields are
20:11rising. Um, and we are also seeing that
20:14prices for junk bonds are rising
20:17massively. So, CoreWeave, for example,
20:20had something like 15%bond yields. That
20:23means the market is reacting to this
20:25investment boom right now by increasing
20:27capital costs because there is so much
20:29demand for capital again, interest
20:31rates are rising. And it will become
20:35incredibly difficult to continue
20:36justifying all the expenditures as
20:38capital becomes more expensive. And the
20:42same thing was what pulled the plug in
20:442000. In 2000, the oil price rose and
20:49interest rates rose, and yes, when
20:51financing conditions become tighter,
20:53meaning there isn't enough capital to
20:56continue financing the expansion, then
20:58such a bubble collapses. And we
21:01currently do not have a bubble like in
21:032000. In 2000, there were no profits.
21:062000 was a bit like crypto in 2017.
21:09There was zero substance there; there
21:10was only hype, no profits. Now profits
21:13are being made, but the profits could
21:15indeed collapse. And that is what
21:17people don't understand. We have, uh,
21:20profits that are currently only this
21:23high because there is so much demand,
21:25so much computing power for the models.
21:29Then so many new data centers are being
21:31built, then those stocks that supply
21:33all the components for these data
21:35centers are making massive profits
21:37right now. But the moment the demand
21:41breaks away, because, for example, a
21:42company runs out of money or because
21:44they aren't profitable or whatever. At
21:47that moment, the plug is pulled. The
21:50higher interest rates rise, the more
21:52likely this all becomes, when we look
21:54at liquidity and the economic cycle. I
21:57do this in my investor group once a
21:59week in detail, so in my investor group
22:01, the Pure Alpha group, we go really
22:03deep into it once a week in the Friday
22:06live update. I'll keep it very brief
22:08now. Uh, we simply see that while
22:11global liquidity is still very high, so
22:13we still have a lot of liquidity in the
22:16system, the momentum, the growth, as
22:18you can see down here, is completely
22:21leaving. And here you can see, above
22:23all—and that is always what interests
22:26me—do we have surplus liquidity?
22:28Meaning, do we have more liquidity than
22:30economic growth? And currently, there
22:34is no surplus liquidity, and that's why
22:37I don't see a major risk-on phase right
22:40now, just a selective risk-on phase
22:42into the stocks that are currently
22:45benefiting the most. More on that later
22:49, um, if you want to know. So, what
22:53happens next? That is the relevant
22:54question now, and sign up for the
22:56webinar, too. I'll say more about that
22:57in a moment. On November. 12th at 7 PM,
23:01I will really go into detail about
23:03which stocks represent the next stage
23:06of the AI trade for me, but we
23:09currently have too little liquidity to
23:12justify a broad risk-on spectrum in the
23:15market. And the important issue is that
23:18oil prices fall and that interest rates
23:20fall. And the decisive factor is, uh,
23:23no, the oil price has risen and
23:24interest rates have risen. what we
23:26would need now for us to get massive
23:29risk-on. Such a final phase, and I can
23:32imagine that we get that now, such a
23:34really intense final phase, the oil
23:36price would have to fall and interest
23:38rates would have to fall, or the Fed
23:40would have to pump in liquidity, then
23:41the AI trade would continue for another
23:43one to two years. That is what we have
23:47to see now. So, that means if interest
23:51rates fall and oil falls, the cycle
23:53will extend, and if interest rates keep
23:56rising and oil keeps rising, we will
23:58get a recession. That is very, very
24:01important. So, uh, if you want to know
24:04what I'm buying now, we'll go through a
24:06few historical examples in a moment. On
24:08the 12th. November at 7 PM, 90 minutes,
24:13100%free. The link is down there in the
24:16video description. Sign up. I look
24:18forward to seeing you. So, it's always
24:21the same script. You have a
24:23breakthrough. Technology, AI, ChatGPT.
24:29Wow, euphoria, everything goes up. Then
24:34you have cheap capital, too much is
24:36invested. At some point, reality sets
24:41in. You can't really change everything
24:44with AI yet. You’ve pushed
24:47expectations above reality and then,
24:50when financing conditions tighten and
24:53there's not enough capital left, the
24:55crash comes. And, uh, mostly there's
24:59this valley of tears, and in this
25:01valley of tears, all the stocks that
25:04were initially hyped lose. Currently,
25:08those are the infrastructure providers,
25:10and the stocks that are really
25:12increasing their productivity with the
25:15new technology are the ones that win.
25:19And, uh, it was the same with
25:21electricity. When electricity was being
25:24built, invented in 1882, then installed
25:27and so on. And you see exactly the same
25:30thing again. Those, uh, who built the
25:34electricity were the ones who went
25:37through this cycle. Those who had
25:41better business models thanks to cheap
25:43electricity are the ones who turned it
25:46into profit and increased their margins
25:48. The same with radio, uh, mobile
25:50communication. The network builders
25:53paid the bill, uh, built the whole
25:55thing, but who made the money with
25:58mobile later on—those were others.
26:02And with the internet, it was primarily
26:04like this with the cloud, right? That
26:08those like AWS, who build the whole
26:09infrastructure, who sell all the
26:11computing power and so on, they earned
26:13all the money. We see here that AWS
26:17revenue in 2015 was still 8 billion,
26:20now it’s 128 billion, and the profit
26:23has expanded massively, and, uh, the
26:26biggest beneficiaries are likely Google
26:29, Microsoft, Amazon, and SpaceX, who
26:32profit most from selling computing
26:34power. Even with the railroad, the idea
26:39was right, just the price was wrong.
26:43They just invested way too much, and
26:46all the companies that built the
26:48railroads went bankrupt. But the fact
26:52that the railroad existed created
26:53economic growth because you had
26:55efficiency gains. You could move goods
26:59faster and so on, but the railroad
27:01stocks, the rail manufacturers or
27:03builders, they went through the
27:06boom-and-bust cycle, were hyped, and
27:08then at some point, it all collapsed.
27:13And if we compare the whole thing now,
27:15we are comparing the railroad era and
27:18AI today. You have the breakthrough,
27:20the first railway line or the internet,
27:23and for AI, it’s ChatGPT. Then comes
27:26the euphoria, that was the railroad
27:29stock hype back then, and the internet
27:31bubble hype. And currently, we are in
27:35that phase of euphoria. Then comes the
27:39phase where there is too much
27:41investment. Overbuilding—too much is
27:44invested, just like in the railroad
27:46days. With the dot-com bubble, there
27:48was also too much investment. You can't
27:51always compare it directly to GDP, but
27:53it’s about investing more than is
27:54needed because profitability isn't
27:56there yet. And yes, the crash follows.
28:01For railroads, it was almost -60%. For
28:05the dot-com bubble, it was nearly -80%,
28:08and I think in the internet age,
28:10semiconductor stocks—not the NASDAQ,
28:13but the semiconductor stocks—will
28:15likely lose 50, 60, 70, or 80%.
28:19Tactically, I actually find them
28:20interesting because they are oversold
28:22in the short term, because we will
28:23likely see a year-end rally, and so on.
28:25But looking ahead to the next few years
28:28, I would be super, super careful with
28:30all these AI infrastructure and
28:32semiconductor stocks. They don't belong
28:34in a long-term portfolio. Who benefited
28:37from that? With the railroad, it was
28:39the industry and commerce. In the
28:41dot-com crisis, it was mainly Google
28:42and Amazon. And who could benefit from
28:45AI? Yes, those are the stocks that
28:48actually apply it. As I said, in the
28:50webinar on the 12th. I will go into
28:53that in very, very, very great detail.
28:55Now, here is my most important summary,
28:57and then we'll look at a few ideas. I
28:59am not bearish in the short term. Very
29:02importantly, I think we will see a
29:04year-end rally. There is currently a
29:06lot of pessimism in the market. Uh, we
29:08have positive mid-term seasonality.
29:11That's why I can very well imagine
29:13we'll see a year-end rally. I am
29:15positioned accordingly, but I don't
29:17think we'll have enough momentum to
29:20reach new all-time highs. I think we'll
29:23have a strong move, then everyone will
29:25get super bullish again, and that will
29:27likely be sold off. As I said, on the
29:2912th of. November, in the webinar, I
29:32will address exactly that. For me,
29:34there are two groups that are
29:35interesting right now. First, the
29:38scarcity winners—where there isn't
29:39enough supply, but demand is still
29:41strong because, for now, demand remains
29:43strong. It's just that the next phase
29:46will be when demand collapses, but for
29:48now, it's strong. And uh while the
29:51first phase was the whole topic of GPUs
29:54, Nvidia and so on, the second phase
29:57was memory. Um Micron is one and so on.
30:00And the third phase, in my opinion,
30:02will be the topic of energy. We simply
30:05see that all the energy companies
30:08currently don't have enough um capacity
30:10to bring the data centers online. But
30:14there are also incredibly exciting
30:15opportunities in the field of photonics
30:17um. As I said, these are not buy
30:19recommendations, these are stocks that
30:21you can take a look at. Um, whether one
30:23still wants to play this phase 1 of
30:25scarcity winners, I have to be honest,
30:27everyone has to decide for themselves,
30:29because phase 2 of those companies that
30:31then increase their margins and such
30:33with AI will be much, much more
30:35interesting. Those will be the
30:38companies that earn more money by using
30:41AI, and potential winners are also
30:43listed here. Interesting, for example,
30:48was an Adobe, which I traded because it
30:50had dropped massively, but improved its
30:52numbers, although for me Adobe will
30:54actually be more of an AI loser in the
30:56long run. Um, in conclusion, three
31:00sentences for you to take away. I know
31:02it was a long video, it was an
31:03intensive video. Feel free to give me
31:05feedback. You said you wanted more long
31:07videos, feel free to give me feedback.
31:08I want to deliver the maximum added
31:10value for you here. The AI revolution
31:12is coming, no question about it. But
31:15what was also said about the French
31:17Revolution, the revolution eats its
31:18children, right? So uh it could eat its
31:22own pioneers. I don't believe that
31:25OpenAI and Anthropic will be the
31:27winners, but rather those who will use
31:29OpenAI and Anthropic profitably. And if
31:33you want to see my stocks with which I
31:36am implementing all of this, then come
31:39by. On November 12th. At 7 PM live. I
31:42will go into it in great detail. The
31:44link for the webinar is down in the
31:46video description.

About this transcript

This page contains the full transcript of Der KI-Hype hat die Realität überholt! by Balthasar Becker, generated from the public captions YouTube serves with the video. The transcript has 4,720 words across 718 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.