Tiger ETF team leader Lim Jong-wook is here again. Thank you for coming. It’s nice to see you. AI, AI, AI... Recently, however, the sentiment seems to be changing a bit. After the earnings reports from big tech companies, there was some volatility, especially following Nvidia's earnings announcement. That's right. At this point, people are starting to have many concerns. How should we approach AI investments? What direction should we take? Today, we will look into ETFs related to AI. Yes, they are all a bit different; they have different characteristics. In a broad sense, AI investment is correct, but if we previously focused on hardware, semiconductors, and big tech, now we need to pay attention to the underlying stocks that are essential for building AI infrastructure. It would be beneficial to approach it from that perspective. While this point may be somewhat controversial, big tech...

Companies are now considering whether to adjust the pace of their investments in data centers, including GPUs. It’s important to address this point first. I don’t think anyone would deny the investment from big tech companies. However, given the current volatility, there might be concerns about whether to continue holding onto these investments. Of course, from a long-term perspective, such worries may not be very significant, but for investors, the immediate market fluctuations inevitably raise these concerns. 

Nevertheless, one clear point is that when investing in big tech, it's important not only to consider how much they are investing but also to look at the companies that are actually generating profits. We also need to pay attention to the infrastructure necessary for AI to function properly. In fact, related investments are expected to continue increasing.

These companies have plans and are actually demonstrating those aspects. Companies like Alphabet, Amazon, Meta, and Microsoft are making capital expenditures that may not yield immediate profits but are aimed at future gains. They are continually increasing their investments, including in data centers. As you can see in the charts, these investments are consistently rising. Just this year alone, the four big tech companies have executed capital expenditures exceeding $100 billion, which represents an increase of over 50% compared to the same period last year. These companies have plans to continue executing investments of over $1 trillion in the future. It's well-known that cloud infrastructure is essential for AI to operate, and for the cloud to function, data centers are also necessary.

This is a part that is essential to take away, and it can be seen primarily as an investment related to data centers. If we look at the comments made by CEOs of big tech companies, for example, the CEO of Google mentioned that during times like these, rather than making a small investment in AI, taking on more risk is much better. When investments are too low, the risk increases, as the saying goes, too much of a good thing can be bad. However, this is quite the opposite. It is better to invest more rather than less, as having a larger portion in that area carries less risk than having a smaller investment. While they have adjusted their pace, they continue to invest aggressively in data centers because once you fall behind, it can be the end. This is a winner-takes-all structure.

Commercial activities may have an impact on short-term profitability, but for now, they are willing to endure these challenges and continue investing. Mark Zuckerberg mentioned that at this point, it is better to take the risk of continuing to invest rather than investing too late. Amazon, Microsoft, and others have made similar statements, which suggests that we can anticipate that big tech companies will continue to increase their related investments. This is truly an AI investment war; if you fall behind even a little, you could end up at zero. However, from an investor's perspective, even if one thinks that this might not be an AI bubble, the current volatility can be concerning. What should one focus on? First of all, if we look at investment assets, then...

Rather than focusing on individual stocks, it seems that approaching investments through portfolios like ETFs is a more viable option. When we look into specific investment targets, investing in big tech is certainly valid. However, when we approach big tech, we should pay more attention to companies that can actually make money through AI. We need to examine their financial indicators to determine whether they are genuinely profiting from AI, rather than just claiming to invest in AI or utilize it.

Companies like Google and Meta might be generating revenue in ways we don't fully understand yet, but we should identify where companies like Google, Meta, and Amazon are actually spending money and generating sales. We need to look for those concrete instances. Additionally, there will be various infrastructures required for AI to function effectively.

Significant investments are being made, so we should pay attention to that. AI is now a mega trend, and if we look at its development, the initial stage will require the establishment of hardware. This is why the semiconductor industry has benefited. Once the hardware is set up, we can expect the industry to develop into software and services. Currently, we are still in the early stages, and from a hardware perspective, semiconductors have continuously benefited since last year. If we take a closer look at this hardware, we can see that it includes semiconductors, servers, and further extends to networking, storage, and eventually devices. It is expected to evolve in this manner. Observing the investment plans related to cloud services from various companies, we can see a consistent upward trend.

To operate data centers, a significant amount of electricity is required, which is often highlighted in the news. The demand for high-voltage cable contracts continues to rise. Currently, as we are in a hardware-focused era, while semiconductor companies are crucial, the operation of these components necessitates cloud services. For cloud services to function, data centers are essential, and to maintain data centers, various resources, particularly electricity, are vital. Therefore, we need to consider all these aspects collectively. This can broadly be referred to as AI infrastructure, but the scope of infrastructure is quite extensive, and there are likely numerous related companies. If you could provide a more detailed explanation of this, it would be appreciated.

좋을 것 같은데, 크게 여러 가지가 있겠지만 세 가지로 나눌 수 있을 것 같아요. 첫 번째가 데이터 센터, 그리고 말씀드린 대로 전력 인프라, 예, 그리고 어쨌든 이런 것들을 만들기 위한 원자재로 크게 세 가지 카테고리로 구분할 수 있을 것 같습니다. 데이터 센터 인프라를 좀 더 세부적으로 보면, 워낙 열이 많이 나기 때문에 냉각과 관련된 부분들, 그리고 전력에 대한 부분들, 자동화에 대한 부분들이 필요할 거고요. 전력 인프라는 전력 공급을 위한 여러 가지, 전력을 만들어내는 것부터 송전하는 것, 배전하는 것까지 여러 가지가 포함될 거고요. 원자재 같은 경우는 전기를 생산하기 위해서 요즘은 원자력이 다시 주목받고 있는 상황이기 때문에 우라늄이나 전선에 들어가는 구리 같은 원자재를 저희가 생각할 수 있을 것 같습니다. 그런데 사실 우리가 데이터 센터에 대한 수요가 엄청나게 늘어나고 있다는

I have heard a lot about this story. Since the emergence of GPT, could you show me which companies are related to the data center industry and how much they are growing? Data centers have actually been a focus even before AI, as most companies have shifted from managing their own servers to utilizing cloud services. With the significant increase in data traffic, the demand for data centers has been consistent for the past 20 years. However, with the advent of generative AI, we can see a renewed boost in this area. 

In the United States, investments related to data centers are continuously increasing, and the vacancy rates are steadily decreasing. Yes, the vacancy rates for data centers are dropping, and rental prices are rising. Therefore, there is an expectation that investments in data center infrastructure will also increase. While there have always been many data centers, especially now...

The data center for AI requires a significant amount of technical expertise. It is not simply a matter of setting up servers; various systems need to be properly maintained. For instance, heat management is crucial because a lot of heat is generated, so cooling systems are very important. Additionally, power supply must be stable; if it gets interrupted, the servers will go down. Networking and storage servers are also essential components. Representative companies in this field include Arista Networks, a networking equipment provider, Dell Technologies, which specializes in AI servers, and Vertiv Holdings, a supplier of cooling solutions. Many investors who are interested in U.S. stocks might be investing in these individual companies.

There is a possibility that everything could be consolidated into one ETF. I would like to examine this in more detail. Additionally, while discussing data centers, one topic that frequently comes up is power consumption. Recently, I asked ChatGPT a question about how much power it consumes. It turns out that the power used is roughly equivalent to keeping a small LED bulb on each time a question is asked. Furthermore, when we generate images, the power consumption is significantly higher. I would appreciate it if you could explore how much power is consumed in the data center sector. Everyone is aware that electricity usage is high, but let me present some specific figures. For example, when you perform a simple search on Google, it consumes about 0.3 watts. However, when using ChatGPT or conducting an AI search through Google, the power consumption increases by several times.

As the demand for AI increases, it is currently consuming an enormous amount of electricity, particularly with Google's AI search, which operates at 6.9 watts. This means that computations need to be performed continuously. Ultimately, we are discussing the shift towards on-device AI because it is much more efficient and consumes less power. We believe that AI will eventually run on individual devices. 

From the current perspective, AI systems consume a significant amount of electricity to operate. When viewed from the semiconductor side, Nvidia's semiconductors, such as the A100, do not consume much power initially, but as the applications related to AI grow, the power consumption increases dramatically. This trend is evident in the data presented on the right side of the table, where we can see that as we progress to B20000, the power consumption inevitably rises.

Exponentially, the core related to data centers can be said to be power. If we look at the actual power usage forecasts, Goldman Sachs predicted that in 2019, it was 200 watts, and by 2030, it would increase by more than five times. In terms of the overall global electricity consumption, data centers accounted for a little over 1% in 2022, but by 2030, this is expected to rise to nearly 10%. By 2030, the electricity consumption of data centers worldwide will surpass that of the entire reason. This is an enormous proportion. Recently, stocks related to our country's power infrastructure surged after Elon Musk raised the issue on X, highlighting the importance of securing transformer supply. As a result, stocks related to transformers in our country also saw a significant increase.

To ensure a stable power supply, we need to discuss the related value chain and companies involved. Simply put, we need power plants to generate electricity, substations to convert the generated electricity for general use, transmission facilities to carry the electricity, and distribution systems to allocate it. If we highlight a few representative companies, we can mention Constellation Energy, which produces nuclear and renewable energy sources; Eaton, which creates systems related to electricity; and Schneider Electric, a provider of energy management solutions. These are some of the key companies in the sector. We have companies that produce electricity, manage systems, and provide energy management solutions, all of which are essential for data centers and the related infrastructure. Naturally, there are also raw materials that are core to this infrastructure. Therefore, when we consider AI infrastructure-related ETFs, we need to look at the entire value chain down to the bottom level.

There are some key points that seem important. I think raw materials will be crucial, and there are various power generation sources for producing electricity. However, nuclear power has its own set of issues, and while we need to consider environmentally friendly options, the existing power facilities have many problems as well. Especially, big tech companies are claiming they are going green, but they are still relying on fossil fuels. In the past, there was a lot of debate about whether nuclear energy is environmentally friendly or not, but it seems to be gaining attention again as a clean energy source. For instance, Microsoft has invested in a small nuclear reactor company called TerraPower. We don't know how things will turn out in the future, but as of now, nuclear power is almost the only alternative that can handle the current demand for electricity.

The reason why nuclear energy is receiving so much attention is that it is becoming increasingly difficult to manage. As of the end of 2023, there are about 430 operational nuclear reactors worldwide, and more than 60 new ones are currently under construction. Countries like China and several others are continuously adding more nuclear power plants, with plans to establish over 160 new reactors in the next 15 years. Therefore, it is easy to predict that as the number of nuclear plants increases, the demand for uranium will also rise. Uranium is a core raw material that cannot be overlooked. The second raw material we can consider is copper. Copper is widely used in electrical wiring, and it is crucial for electrical equipment. This is another area that we need to pay attention to.

You can think of this as related to AI infrastructure, and therefore, the demand for copper is expected to continue to increase. By 2030, there might be a slight decrease, but as data centers continue to grow, the demand for copper will keep rising. So, in terms of raw materials, uranium and copper will be very important points. I understand that all of this is necessary; big tech companies cannot fall behind, so it's clear that they are continually investing in data centers. Additionally, companies that manage energy and those that produce energy, as well as the need for uranium and copper, are all understood. However, for those who invest, as many of you may know, the market is highly volatile. Copper is no exception, and this is something that must be endured. The volatility is too high for individual companies or individual raw materials. Therefore, the important thing is the concept of bundling these investments together.

To manage volatility, it is naturally meaningful to approach it through a portfolio. Moreover, the AI infrastructure is so diverse, as you mentioned, that it is difficult for us to predict where we will benefit the most. However, we know that everything along the value chain is important, which is why it seems even more crucial to invest in packages. This applies to all new growth themes, but in the early stages of industrial growth, it is hard for us to gauge who will be the winner. Therefore, it would be beneficial to hedge the risks associated with individual companies while approaching ETFs as a portfolio. So, let me introduce a product that will be listed on September 10th: the Tiger Global AI Infrastructure Active ETF. The Tiger Global AI Infrastructure Active ETF will invest broadly across the entire value chain related to AI infrastructure, including data centers, power, and raw materials, as I mentioned earlier.

It is a product, and naturally, the target investment countries will be the global market. What is special is that since it is actively managed, unlike the U.S. market, it is still quite impossible for our country to have pure active management. We cannot simply create an ETF and buy it on our own. While we cannot operate like a typical active fund, there are differences compared to general passive ETFs. A passive ETF sets a benchmark index and replicates it 100%. However, for an active ETF, as long as the correlation with the benchmark index is above 0.7, it does not have to replicate it exactly. This allows for some flexibility in stock selection based on the manager's judgment. In this regard, it can demonstrate even more advantages in themes, as the industry is changing dynamically, allowing for more freedom in stock selection.

There is a bit more room for flexibility. Yes, passive products do experience some adjustments during rebalancing, but compared to active ETFs, there is slightly more inconvenience. Therefore, we can also consider the advantages of managing actively. So, what is the benchmark index? Our Mirae Asset benchmark was created by our affiliate, and it is the Mirae Asset AI Infrastructure Index. You can think of it as being the same as the FTF name. While it tracks this index, it does not actively replace components; rather, it serves as a reference benchmark. Investors can think of it as aiming to outperform this benchmark. So, when we look at the actual components included, we can see...

You should also keep in mind that we can obviously discuss benchmarks now. Earlier, you briefly mentioned a few companies related to data center power and raw materials. What are some representative companies and how are they structured? Currently, the recent portfolio, as you can see on the screen, can serve as a reference. It seems that investments are being made evenly across various data center power raw materials. As you can see, while the U.S. has the highest share, there is a diversification across several countries. There are also many well-known names, including overseas companies that our retail investors like. Some of us are diligently listening to conference calls among these companies. So, AI investment is something that really needs to be viewed in the long term.

Isn't it true that AI infrastructure is also part of this, and now there are actual companies making money in this space? Lastly, if you have any advice, please share that as well. The truth is, even if it's not strictly from an investment perspective, we are already significantly exposed to AI, and we can't avoid it. Of course, when considering investments, especially when thinking about potential profits, we find it difficult to think of stocks other than those related to big tech in AI. In the past, we focused on hardware, particularly semiconductors, but now the next step is identifying AI stocks that are generating revenue. For this to be truly implemented and widely adopted, we need to consider what infrastructure is required. We can deduce that NexThing will be related to that infrastructure aspect.

Investing in those areas, along with investments in big tech, could help enrich your investment portfolio. I believe this is an important consideration. AI is great, NVIDIA is great, and big tech is great, but we should also focus on the infrastructure companies that are actually spending money and generating profits. This includes data centers and the raw materials that power AI infrastructure. Today, we learned about the Tiger Global AI Active ETF. Thank you, Team Leader Lim Jong-wook, for the great explanation. Thank you.

"AI 거품론 불안해요" 진짜 돈 버는 AI 기업들에 투자하는 법 / 미래에셋자산운용 TIGER 글로벌AI인프라액티브 ETF

"지식과 자산의 복리효과" 그 길을 복길이 여러분과 함께 걷습니다.
( 업로드 : 오전 11시45분, 오후 4시, 오후 7시 )
제휴 문의 : ystreet2021@gmail.com

인터넷 언론사 제호 : 와이스트릿 / 등록번호 : 서울 아53714 / 등록일자 : 2021.05.24
주소 : 서울시 영등포구 국제금융로8길 27-9 여의도 동북빌딩 910호
대표번호 : 02-6949-1701 / 팩스 : 02-6949-1702
이메일 : ystreet2021@gmail.com
발행인 : 이대호
Copyright © 와이스트릿  YSTREET ALL RIGHTS RESERVED. 와이스트릿의 모든 콘텐츠는 저작권법의 보호를 받으며, 무단 전재·복사·배포 등을 금지합니다.


"AI 거품론 불안해요" 진짜 돈 버는 AI 기업들에 투자하는 법 / 미래에셋자산운용 TIGER 글로벌AI인프라액티브 ETF

"AI 거품론 불안해요" 진짜 돈 버는 AI 기업들에 투자하는 법 / 미래에셋자산운용 TIGER 글로벌AI인프라액티브 ETF

스크립트