今年收益 97%，滙豐晉信陳平：AI 週期至少 10 年，不認可 AI 泡沫論，TMT 仍是買點，看好 AI 硬件，最看好光模塊/光通信

On December 23rd, at the ninth Value Investment Forum held by our company, Chen Ping from HSBC Jintrust shared his views and judgments on the current technology sector, particularly in the field of artificial intelligence (AI).

The representative of the investment workbook summarized the key points as follows:

1. We believe the current market atmosphere is relatively good, with a neutral to optimistic outlook. The CSI 300 is currently roughly back to the valuation center, and the risk compensation is still in the positive range, making it attractive.

The situation for growth stocks represented by TMT is similar, with valuations slightly below the central valuation, and risk compensation is also at historical highs, around one standard deviation, still considered a buying point.

1. We are more optimistic about growth stocks, especially those represented by AI. Promising industries include: AI, semiconductors, the internet, pharmaceuticals, and commercial aerospace.

Among them, the current pharmaceutical (innovative drugs) market may just be beginning, with greater potential in the future.

1. Overall, we are optimistic about AI hardware, particularly "optics" (optical modules/optical communication), which is growing significantly faster than the overall market.

2. In terms of AI applications, we are relatively optimistic but prefer the AI applications of large companies. For many small and medium-sized internet or computer companies, their competitiveness in the AI era may be questionable and needs to be examined individually.

3. We do not agree with the theory of an AI bubble. Although current revenues are insufficient to fully cover capital expenditures, this is a norm in the early stages of emerging industries, often requiring upfront investment.

4. We tend to believe that AI will have a relatively long cycle, at least at the level of smartphones (ten years), and may even reach the level of the Fourth Industrial Revolution (decades).

5. The core targets in AI currently have low valuations and will remain in a high growth phase for the next few years.

Chen Ping, a fund manager at HSBC Jintrust, has 13 years of experience in the securities industry, with over 10 years focused on technology sector investment management.

Since 2012, he has been deeply involved in the TMT field, experiencing the startup phase of the smartphone supply chain and fully traversing multiple growth stock cycles. In July 2015, he officially took over the HSBC Jintrust Technology Pioneer Fund.

His professional background also includes previous positions as a researcher at Nanjing Securities Research Institute, Guojin Securities Research Institute, and HSBC Jintrust Fund Management Co., Ltd.

Chen Ping currently manages a total fund size of 579 million yuan, with only one product, the "HSBC Jintrust Technology Pioneer Stock."

This fund has performed outstandingly this year, with a return rate of 96.51%, ranking 14th among 969 similar funds; the return rate over the past six months has also reached 87.05%.

Although the market has fluctuated recently, the return rate over the past three months has fallen to 10.86%, but it still ranks 55th among 1040 similar products, maintaining strong relative competitiveness.

The fund mainly invests in technology growth stocks.

Regarding future operational strategies, Chen Ping pointed out in the third quarterly report that after the market experienced a short-term decline due to tariff impacts followed by a rebound to new highs, the valuation attractiveness of many assets has decreased. However, some assets represented by core AI growth stocks are still in a phase of low valuation and fast growth, and we remain relatively optimistic about the performance of these assets over a longer time frame.

Our future operational strategy for the product is to maintain a relatively high position, primarily investing in growth industries. Among growth industries, we are relatively optimistic about AI-related (hardware, applications), electronics (semiconductors, consumer electronics), etc.

The following are the highlights organized by the investment operations class representative (WeChat ID: touzizuoyeben), shared with everyone:

CSI 300 Returns to Valuation Mean, Risk Compensation in Positive Range, Attractive

Hello everyone, good afternoon. The topic I am sharing is "AI in Progress, Technology Remains the Main Line."

Overall, we believe the current market atmosphere is good, with a neutral to slightly optimistic outlook.

Let's first look at the current position of the market.

The CSI 300 is currently roughly back to the valuation mean, and risk compensation remains in the positive range, making it attractive.

TMT Still a Buying Point

Even more attractive are growth stocks, such as the ChiNext Index. Its valuation is still below the historical mean, while risk compensation is near one standard deviation above the mean, historically indicating good timing for buying and selling; it is still in an attractive range.

The situation for growth stocks represented by TMT is similar, with valuations slightly below the mean and risk compensation at historical highs, around one standard deviation, still considered a buying point.

The pharmaceutical sector is also at a high level of risk compensation. After the market rise, valuations have returned to the mean, while risk compensation remains attractive.

Looking at the market from the three DCF elements: performance (numerator), risk-free rate, and risk preference.

The latest situation can be summarized as follows: we expect the economy to be stable next year, showing a moderate recovery trend, with overall stable performance from listed companies, contributing slightly positively on the numerator side.

Regarding the risk-free rate, we expect the U.S. to be in a rate-cutting cycle next year, likely with two rate cuts; China's risk-free rate is expected to remain stable overall.

In terms of risk preference, the current market risk preference is good. The phenomenon of deposit migration and capital inflow into the stock market is gradually occurring, with capital entering the market still in the early stages. We expect risk preference to remain stable.

Considering the above factors, we believe the overall market presents a neutral to slightly bullish pattern. This is our judgment on the overall trend.

Optimistic About Five Industries

Specifically regarding sectors, we are more optimistic about growth stocks, especially those represented by AI. In a period of moderate economic growth, growth stocks highlight their performance advantages with faster earnings growth, attracting capital attention.

Therefore, growth stocks possess both performance and risk preference advantages. If combined with a declining interest rate cycle, their performance is usually even better.

We are relatively more optimistic about growth stocks. That is: the overall trend is neutral to optimistic, with a focus on growth in sectors Promising industries include: AI (to be discussed later), semiconductors, the internet, pharmaceuticals, and commercial aerospace.

In the semiconductor field, the trends of China and the U.S. were once synchronized, but have since diverged, with the Philadelphia Semiconductor Index continuing to strengthen. Starting this year, Chinese semiconductors have begun to catch up, with AI being the core driving force.

Previously, AI boosted U.S. semiconductors with little connection to China; starting this year, cloud computing power (such as domestic GPUs) and edge-side semiconductors have brought new demand, driving semiconductors into an upward cycle, which we remain optimistic about. The long-term logic of domestic substitution still holds.

Consumer electronics currently have low valuations, and AI may bring opportunities for valuation increases, whether in AI smartphones, AI PCs, or other edge-side devices. The combination of low valuations and new opportunities may create a "double hit" in terms of valuation and performance.

The internet sector is overall not expensive, has strong core competitiveness, and possesses core scenarios for AI implementation. The stock price trends of core internet companies in China and the U.S. were once similar, but have diverged since the emergence of AI.

AI has already contributed to North American giants, and we can also see this in Chinese internet companies, for example, the improvement in advertising effectiveness driving growth rates above the industry average. Under the influence of AI, if valuations are low, there is room for improvement. Their performance is also relatively stable.

The innovative drug market is just beginning, with greater future potential

We are also optimistic about pharmaceuticals (innovative drugs), believing that the current market may just be the beginning, with greater potential ahead.

Commercial aerospace is not being focused on due to recent speculation. We have been researching it for over two years and have relevant holdings. Its future potential is vast.

One favorable factor is that Musk (SpaceX) has demonstrated a feasible model and the Starlink constellation, with the business model validated, and we are in a catch-up phase. This is different from pursuing unverified fields.

For example, the Long March rocket recovery test failed on the morning of the 23rd, but this is normal; new industries often start with attempts.

Optimistic about AI hardware, especially optics

Next, we will focus on AI, which everyone is concerned about. This year, the AI sector has performed strongly, with many stocks doubling. How should we understand AI and its future outlook?

This year, AI has entered a positive cycle where model capability enhancement and application implementation mutually promote each other. This is reflected in the continuous growth of token usage (both U.S. and Chinese giants), while revenue is beginning to materialize. For example, OpenAI's annualized revenue may reach $20 billion by the end of the year, with expectations of $50 billion next year and $100 billion the year after, showing explosive growth.

Not limited to OpenAI, many companies' AI-related revenues are also rapidly increasing. Some giants, although not separately listing AI revenue, have already seen effects in their businesses. AI is entering a positive revenue cycle.

Giants continue to increase AI capital expenditures. This year, global AI capital expenditures are about $500 billion, and may reach $700 billion next year. Jensen Huang of NVIDIA has predicted that by 2030, the scale could reach $3-4 trillion. Based on this estimate, the compound growth rate in the coming years may approach 50%, indicating sustained high growth over the next five years.

Against this backdrop of high capital expenditures, multiple sub-sectors may experience even faster growth First of all, the proportion of capital expenditure allocated to core computing, communication, and other areas closely related to AI is becoming increasingly high. In contrast, the share allocated to infrastructure construction (such as building data centers) is decreasing. Therefore, the growth rate of this part (core computing and communication) should be higher than the overall level.

Furthermore, among all these core devices, you will find that the proportion of GPUs is actually declining. Why is that? Because GPUs can follow Moore's Law for iteration—Moore's Law is a very significant effect, with astonishing improvements in both performance and cost reduction. According to this law, performance doubles or costs are halved approximately every 18 months, right? So its evolution speed is extremely fast.

However, for most other devices, we can categorize the part that follows Moore's Law as "electronic" products, while the rest belong more to the "physical world" of tangible components. The progress in these components, whether in terms of cost reduction or performance improvement, is generally relatively slow. For this reason, the growth rate in segmented fields, including optical modules, PCBs, power supplies, etc., will actually be faster than the overall capital expenditure growth rate.

Among these fields, we are relatively most optimistic about "optics" (optical communication/optical modules), whose growth rate is significantly higher than the overall.

For example, if the proportion of ASICs increases in the future, the proportion of optics will also rise accordingly. This is because typically, the single-chip capability of ASICs is not as strong as that of NVIDIA's GPUs, so more connections are needed to interconnect these chips to achieve or approach the performance of NVIDIA GPU clusters. Therefore, the increase in the proportion of ASICs will drive demand for optical interconnects.

This is why, since the beginning of this year, you have seen strong stock performance in these areas (such as optical modules), which is actually backed by very strong fundamentals.

The demand for optical modules continues to be revised upward, and it is still in a relatively early stage, with continuous technological advancements and ongoing iterations—each iteration represents a new opportunity. Many of the three-letter abbreviations you hear, such as CPO, LPO, etc., are all future technological directions.

In addition, the optical module market we are currently referring to mainly pertains to the Scale-out field. However, especially in the past two months, we have gradually begun to see an increasing demand emerging in the Scale-up market.

Scale-up could be 5-10 times that of Scale-out, with current connections primarily completed by copper and PCBs.

As performance requirements increase, copper and PCBs may become limited, and optics will enter the Scale-up field, further opening up market space.

Therefore, overall, we are optimistic about AI hardware, with a particular focus on "optics."

AI Applications Are Flourishing, but the Landscape Is Uncertain

In terms of AI applications, we are relatively optimistic, but we lean towards the AI applications of large companies.

For many small and medium-sized internet or computer companies, their competitiveness in the AI era may be questionable and needs to be examined one by one

AI Cycle Lasts at Least 10 Years, Rejecting AI Bubble Theory

Responding to several key issues regarding AI.

First, the AI bubble theory: We do not agree. Although current revenues are insufficient to fully cover capital expenditures, this is a norm in the early stages of emerging industries, which often require upfront investments. Revenues are growing rapidly. Cash flow has not yet reached a stage where it cannot support expenditures, and there are various financial tools in North America that can optimize reports, making investments sustainable.

Second, the "funding cycle" question: The entire economy is essentially a funding cycle. In the early stages of AI development, there are relatively few participants, which may make it feel like a "scam," but this is normal. As development progresses, the cycle will continue to expand.

Third, the AI Capex (capital expenditure) cycle is currently experiencing a peak in investment; everything is cyclical, which is correct. We tend to believe that AI will have a relatively long cycle, at least at the level of smartphones (ten years), and it may even reach the level of the Fourth Industrial Revolution (decades). If the cycle lasts more than five years, it can be considered growth, and we are not overly concerned about this issue.

Then, the return of high ROE: Currently, the net profit margins and ROE (which can reach 50%-100%) in segments like optical modules are very high. This industry has technical barriers and continues to improve, which is favorable for leading companies.

Unlike many industries that are overly competitive with rapid profit regression, this is a market dominated by overseas players with participation from China, and the competitive landscape is relatively good, similar to the initial Apple supply chain—expansion has not led to obvious oversupply, and profit levels are maintained.

Currently, the demand for AI computing power continues to explode, and supply is somewhat lagging behind demand. The return of ROE will take a long time, indicating that future profitability is expected to remain stable.

Core AI Targets Still Undervalued, Next Year's PE May Be Below 20 Times

Next, capital clustering: "Clustering" usually refers to individual rationality leading to group irrationality, pushing valuations to unreasonable levels. The current situation is that research shows AI performance is good and valuations are cheap, making capital purchases normal; we have not yet reached a stage of group irrationality.

Core target valuations are still low, and next year's PE may be below 20 times, which cannot be considered severe clustering. For new things, the divergence between bulls and bears is normal, and having a large number of bears is somewhat beneficial.

Because currently, a target with a 20 times PE and an expected compound growth rate of 100% over the next two years makes decision-making relatively easy; if the market unanimously expects a 100 times PE, the difficulty of decision-making actually increases.

AI May Reshape Economic and Employment Landscape

Finally, AI and unemployment: Technological innovation often accompanies unemployment, such as cars replacing coachmen. The impact of AI this time may be broader, as the proportion of the virtual economy increases, and AI excels in the sandbox field. The degree of AI replacement depends on its capabilities and may have a significant impact on employment, especially potentially leading to more structural unemployment, with young people being more significantly affected in the short term.

In the long term, AI may reshape the economic and employment landscape. For example, the cost of call center jobs will be defined by the cost of the AI computing power that replaces them; if the computing power cost is extremely low, that job may disappear AI may also exacerbate the wealth gap if a large amount of labor is replaced by AI, while AI computing power is concentrated in the hands of large capital.

North America is leading in AI investment and exploration, and its tolerance may also be higher; we can observe its development experience.

In summary, we remain optimistic about AI, especially core targets. Current valuations are not high, and it is still in a high growth phase for the next few years.

The overall view is that the market trend is neutral to optimistic, favoring growth stocks, with AI being the most favored among growth stocks. Other mentioned industries are also optimistic.

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

Risk warning and disclaimer

The market has risks, and investment should be cautious. 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 align with their specific circumstances. Investing based on this is at your own risk