Wu Chao said: The two directions he is most optimistic about in 2025

DeepSeek continues to ignite the AI industry circle, accelerating the implementation of AI applications. What changes will it bring to the market?

On the evening of February 11, Wu Chaoze, the director of the research institute at CSC and chief analyst for the TMT industry, shared his views on the market and AI in 2025 with Chen Guo, the chief analyst for strategy research at CSC.

The investment workbook representative summarized the key points as follows:

Wu Chaoze:

1. Major technological cycles generally have a ten-year cycle. The technological changes brought by this round of technology or AI should be similar in magnitude to those ten years ago, but the timing still needs further judgment.

2. Our core view for 2025 is “domestic computing power is ready, AI applications take root,” which also represents the two directions we are most optimistic about.

3. The biggest difference between 2025 and 2024 is that previous investments were mainly concentrated in the TMT field, such as computing power, and AI targets based on TMT, but the keyword for 2025 is opportunities outside of TMT.

4. The difference between 2025 and 2024 is that there will be differentiation. It will no longer be simple concept speculation, but rather whether companies have orders and revenue contributions.

2024 was mainly about concept speculation, while 2025 will focus more on fundamentals.

Chen Guo:

1. Our view on the market for 2025 can be expressed in one sentence: confidence reassessment bull, from liquidity bull to fundamental bull.

2. Next, we will see further liquidity easing, ultimately leading to fundamental improvement, which is also a core factor supporting the continuation of the A-share bull market in 2025.

3. The industry allocation for 2025 still needs to focus on the TMT industry, with the second being AI+, especially China's intelligent manufacturing. Additionally, attention should be paid to industries that are related to the capital market and are in a good industrial cycle.

The following is the essence compiled by the investment workbook representative (WeChat ID: touzizuoyeben) to share with everyone:

Wu Chaoze: Domestic computing power is ready, AI applications take root

Host: When we discuss the bull market, we often say that technological progress will bring about a bull market. So what changes will technology represented by AI+ bring in 2025?

Wu Chaoze: I remember always using an analogy that the mobile internet wave in 2015 left a deep impression, just ten years ago. Looking further back, 2005 was also a wave of the internet. In fact, major technological cycles generally have a ten-year cycle. However, the timing may not necessarily be a completely linear 10 years. Last year, our core view on technology was that new technologies similar to mobile internet were just beginning.

In 2013, I wrote a report titled "Mobile Games: The First Profitable Application of Mobile Internet." At that time, people first saw the practical significance of 3G and smartphones on application scenarios, which would indeed be reflected in financial statements, bringing incremental performance to companies.

The subsequent events are well known; starting from 2014 and 2015, the internet plus everything transformed many traditional industries, from education and technology to finance. Therefore, the current round of technological changes brought by technology or AI should be on a scale similar to ten years ago, but the timing still needs further judgment.

In the past two years, we have seen that since the emergence of ChatGPT at the end of 2022, large models have begun to rise. This is actually more like the initial stage of the underlying hardware and operating systems of mobile internet. So the impact of this round of technology may have just begun.

Host: In the previous round, you wrote "Internet + Everything," and this round has changed to "AI + Everything."

Wu Chaoze: Our core view for 2025 is "Domestic computing power is ready, and AI applications will take root." These eight words also represent the two directions we are most optimistic about.

On one hand, computing power will continue to spiral upwards due to application iteration and the iterative demands of large models. On the other hand, although the computing costs of large models have recently decreased, the demand for computing power based on training may decline, while the demand for computing power for application inference will rise significantly. This is because for applications to truly take root, a necessary condition is that computing costs must decrease. Only then can it transition from the technological or model era to the application era, similar to the situation during the mobile internet era.

In 2013 and 2014, another important change was the significant drop in traffic prices, which made people willing to use it for watching videos and e-commerce. Therefore, one key point to pay attention to in 2025 is the decrease in large model costs, which will further promote AI applications to "take root."

Currently, whether overseas or domestically, most applications are still in non-serious scenarios providing emotional value in the to-c consumer field, such as chat customer service. However, by 2025, as the reliability and traceability of models improve, AI will be widely applied in serious scenarios, such as in the to-b industrial sector.

Host: The application scenarios in the industrial field are more numerous and complex.

Wu Chaoze: Indeed, there were not many industrial application scenarios during the mobile internet era, mainly focused on consumer (to C) consumption internet. But in the AI era, industrial scenarios or enterprise (to B) fields have higher data quality, shorter supply chains, easier closure, and are more likely to achieve cost reduction and efficiency improvement. Therefore, this year, mainly some enterprise-oriented SaaS companies in the United States have benefited.

The keyword for 2025 is opportunities beyond TMT

Host: With strong demand for AI computing power, what opportunities are there in segmented industries in 2025? Wu Chaoze: I believe AI will definitely become the foundation of all industries. From the perspective of investment in the U.S. since 2022, it can be said to some extent that "no investment without AI."

Traditional industries will also undergo significant path changes due to AI technology. For example, in the electric vehicle industry, the first half is electrification, and the second half is intelligence, which reflects the profound impact of AI on the electric vehicle sector. Additionally, this year's Nobel Prizes in Chemistry and Physics are both related to AI.

Therefore, the biggest difference between 2025 and 2024 is that previous investments were mainly concentrated in the TMT sector, such as computing power and AI targets based on TMT, but the keyword for 2025 is opportunities outside of TMT.

The industrial trend changes brought about by AI technology will create significant catalysts or turning points in cost reduction and efficiency improvement, allowing some companies to stand out in their original fields. These companies may not be in the traditional technology sector but will benefit from AI technology.

I believe good targets may lie within these sectors, which is the most important observation perspective for AI investment in 2025. The valuations and performances of these companies are actually more promising than those of companies in the technology sector that have been speculated on for two years.

2025 Domestic Computing Power Will Also Change Significantly, with GPU as the Core

Returning to the issue of computing power, the biggest doubt in the market now is whether the marginal growth rate of computing power can maintain a high level. This is because investors are more concerned about growth rates, or slopes.

For example, the NVIDIA supply chain and domestic computing power have seen the fastest growth rates in the past two years. The first quarter of 2023 marked a significant turning point, as large models emerged at the end of 2022. Since then, the capital expenditures of North American cloud providers have been growing at double-digit rates month-on-month. Domestic computing power will also see significant changes in 2025.

On one hand, large model manufacturers, such as BAT and ByteDance, are entering a new expansion cycle for capital expenditures.

On the other hand, in the past, the underlying capital expenditures for computing power in China mainly came from the three major telecom operators, amounting to about 300 billion annually. However, in 2025, (Chinese) internet companies' capital expenditures may also exceed 300 billion, comparable to those of telecom operators. The sum of these two absolute values has never been seen in history. The core remains GPU, and domestic GPUs will be a focus.

Unlike investing in the NVIDIA supply chain in 2024, we cannot buy the most core GPU companies.

The Rise of Hardware in 2024 Will Continue

Host: What specific sectors can we invest in?

Wu Chaoze: In addition to GPUs, the entire industry chain also includes servers, network equipment, and data centers (IDC). With GPUs, they are assembled into servers, equipped with network devices, and deployed in data centers. This will drive the entire data center's power supply, liquid cooling, PCB, and other segments, with prosperity improving alongside iterations.

Host: I noticed that many stocks in these hardware-related fields you mentioned have already been rising since December 2024. Will this continue into 2025? Wu Chao said: It will definitely continue.

Because the large technology cycle has just begun. The underlying large models are becoming increasingly mature, and their applicability has significantly increased. In the future, if more industries begin to industrialize AI—everyone sees AI, understands AI, and uses AI—the demand will increase exponentially.

In 2024, the focus will be on AI concepts, while in 2025, more attention will be paid to fundamentals

However, 2025 will be different from 2024, as it will diverge. It will no longer be simple concept speculation, but rather whether companies have orders and revenue contributions.

In 2024, the main focus will be on concept speculation, while in 2025, there will be more emphasis on fundamentals. In the hardware sector, many targets are relatively easy to select, even including upstream semiconductors. The upstream of GPUs is different from the mobile internet era; it requires its own operating system, software ecosystem, and advanced process capacity, all of which will be reflected in financial statements.

In 2024, our theme is "a hundred flowers bloom," and we are now entering a more pragmatic stage.

Chen Guo: From liquidity bull to fundamental bull in 2025

Host: Summarize today’s overall view in one sentence.

Chen Guo: In 2025, our view on the market can be summarized as a confidence reassessment bull, transitioning from a liquidity bull to a fundamental bull.

The core of this is that since September 24, the risk appetite in the A-share market has significantly increased. Moreover, the political bureau meeting and the central economic work meeting at the end of 2024 have clearly outlined the tone for 2025, which includes broad monetary policy and fiscal expansion, fully revitalizing the economy, raising price levels, and stabilizing asset prices. Next, we will see further liquidity easing, ultimately leading to fundamental improvements, which is also a core factor supporting the continuation of the A-share bull market in 2025.

From the perspective of cycle theory, 2025 will see changes in fundamentals or economic cycles.

From financial cycles, capacity cycles to inventory cycles, the Chinese economy and corporate profits in 2025 will exhibit characteristics of a cyclical bottoming and upward resonance.

This is specifically reflected in: central fiscal leverage, central bank balance sheet expansion, and corporate supply-side optimization.

Some industries will see supply contraction, while others will begin to see demand bottoming out and recovering. This cyclical resonance will lead to an overall recovery in corporate profits.

From the perspective of specific industries, the number of industries with fundamentals bottoming out and improving will increase in 2025.

In 2024, supported by the old-for-new policy, the domestic fundamentals of industries such as automobiles and home appliances have begun to improve.

In 2025, under the backdrop of broad monetary and fiscal support for total demand and supply adjustments, this improvement will further spread, and we may see recovery in the consumer electronics and machinery industries.

At the same time, due to the boost to the capital market and stabilization of the real estate market, related industries such as finance and real estate will also see improvements. Ultimately, some late-cycle industries, including consumer and service sectors, will stabilize in fundamentals as the overall economy stabilizes.

In summary, in 2024, some industries have already begun to show signs of bottoming out and fundamental recovery. **In 2025, due to policy and industry-specific factors, this improvement will further spread **

It is very important to note that from a macro perspective, the entire technology industry cycle is still ongoing, especially the new productive forces represented by AI+, which may further promote the transformation of China's economic structure and structural opportunities in the capital market.

Industry Allocation Preferred TMT

Host: From an overall strategic perspective, if you were to allocate industries for investors in 2025, how would you allocate them?

Chen Guo: First of all, based on our judgment of the overall industry cycle and market trends, the industry allocation for 2025 should still focus on the TMT industry.

Whether looking at the situation in 2014 and 2015, or the performance of the U.S. stock market in the internet, mobile internet, and this AI wave, the technology industry is unavoidable.

However, in terms of overall ranking, the first area I would focus on is TMT, the second is AI+, especially China's intelligent manufacturing, as manufacturing is a key focus. In addition, industries that are related to the capital market and are in a good industry cycle are also industries and logic that we believe should be prioritized in industry allocation for 2025.

Overall, we believe that in 2025, we should still have a bull market mindset, as the opportunities in China's stock market in 2025 outweigh the risks. Of course, we also believe that investing in great changes is more important than investing in great companies, meaning that in 2025, we need to seize industrial transformations and position ourselves in the most competitive companies.

Source: Investment Workbook Pro Author: Wang Li

For more insights from industry leaders, please follow↓↓↓

Risk Warning and Disclaimer

The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial conditions, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investing based on this is at your own risk