GF's Liu Chenming: Low Probability of Global Bear Market, AI Industry Trend Continues

Recently, Liu Chenming, Chief Strategist and Assistant Director of the Research and Development Center at GF Securities, shared his insights at a Bosera Fund event on the theme of "Can the Bull Market in Non-US Markets Continue? Spring Strategy Outlook for Asset Allocation."

In his view, at the current juncture, facing the situation in Iran, the question that needs to be answered is, can the bull market trend return in the future? Especially for stock markets primarily in the Greater China region.

Alternatively, a more pertinent question is, if the market rebounds in the future, should we take advantage of the rebound to cash out and continuously reduce equity asset positions, or will it return to the previous slow bull trend?

His stance is clearly on the optimistic side. He believes that for equity assets, the major industry trend of artificial intelligence (AI) has not ended. The global market will not enter a major bear market.

Key Quotes:

1. For East Asian markets, including A-shares, the core support remains the major industry trend of artificial intelligence.

2. Usually, interest rate hikes and liquidity contraction severely suppress valuations. However, when the market cannot find other sectors with high prosperity, localized high prosperity will become the main direction for capital concentration.

3. The most important factor in the second quarter is structure; structural allocation is clearly more significant than overall positions.

4. As long as two points are confirmed – that the global economy will not enter a recession in the second half of the year and the AI trend will not collapse – then during the "April Decision" period, it is feasible to identify sectors with structural high prosperity.

5. If one pessimistically believes that the global economy will inevitably decline and enter a bear market, then there is no need for an "April Decision"; instead, one should continuously reduce positions.

6. In April, we should look for industries that are less affected by high oil prices and high interest rates, possess independent potential for localized explosive growth, and can pass on cost pressures downstream.

7. In the past two years, China's AI industry chain has mainly been driven by thematic investment. However, this year is likely to be a breakout phase with a steeper growth slope for token consumption, marking the "Year One" of widespread data center deployment in China.

8. The second quarter presents a rare window of respite for Hong Kong stocks. If global market risk appetite stabilizes, Hong Kong stocks will have an opportunity for valuation repair in the second quarter.

9. For investors looking to hold indices like the Hang Seng Tech, it is still necessary to wait for the emergence of profit elasticity, which means waiting for the recovery of broad Chinese demand.

Presented in the first person; some content has been abridged.

How to View the Iranian Issue

Before this Iranian black swan event occurred, we observed that non-US markets experienced a "double-click" bull market around the Chinese New Year and the following week. European, East Asian, and some Southeast Asian markets all reached new historical highs. After the Chinese New Year, China's A-share Wind All A Index also hit a new high in this rebound phase. However, these trends were interrupted by the Iranian issue.

Therefore, the first sharp question we need to answer is, can the bull market trend return in the future? Especially for stock markets primarily in the Greater China region.

Before answering this question, let's look at the current state of short-term market sentiment to judge the medium-term trend.

There are many indicators for short-term sentiment. While a commonly used indicator cannot determine bull/bear markets or major turning points, it can indicate whether short-term sentiment is overheated or at freezing point. Once this indicator reaches the 100th percentile, it signifies market overheating. In mid-January of this year, it reached the historical 100th percentile, indicating overheated market sentiment. Consequently, thematic investment or sentiment-driven sectors and stocks largely peaked in mid-January.

After a series of developments and the outbreak of war, the sentiment indicator in the past two weeks has fallen back to the 0th percentile. Historically, every time it reached the 0th percentile, it marked a freezing point for market sentiment, and the market would achieve a phase of stabilization.

Recently, there has been some possibility of negotiations between the US and Iran, potentially deferring the most pessimistic scenarios for now.

In the short term, the worst point for risk appetite should have passed.

The Core Support Remains the Major Industry Trend of Artificial Intelligence

A more pertinent question here is, if the market rebounds in the future, should we take advantage of the rebound to cash out and continuously reduce equity asset positions, or will it return to the previous slow bull trend?

Regarding this, especially for East Asian markets, including A-shares, the core support remains the major industry trend of artificial intelligence. The AI industry is driving a significant portion of the economic fundamentals and elasticity in East Asian markets, even spawning many associated thematic investment opportunities. Therefore, the crucial question is whether the major industry trend of AI itself has any issues.

This also brings up macroeconomic concerns about the impact on the AI industry trend. For example, if the US enters a recession, the impact on the earnings and capital expenditures of US tech giants, as well as the pressure from financing costs. Furthermore, the issue of overseas liquidity; if the Federal Reserve does not cut interest rates, what impact will it have on the financing costs of major companies, market liquidity, and valuations, and how will this subsequently affect the entire AI industry trend? These are questions that need discussion.

Let's first consider the recession issue. Will a recession lead to a deterioration of fundamentals for US AI-related companies, or will capital expenditure be unsustainable? In April and May of last year, the market was highly concerned about this, anticipating a risk of recession in the US economy. Consequently, stocks related to overseas computing power and US tech giants experienced a significant adjustment.

In recent roadshow discussions, I've found that this remains a latent concern for many. The perceived logical transmission path is superficially straightforward: war breaks out, oil prices surge, leading to an inflation rebound, potentially causing the Federal Reserve to continue raising rates or refrain from cutting them, ultimately resulting in an economic recession. This logical chain appears simple and direct, but reviewing historical experiences, many aspects warrant further discussion or require consideration in conjunction with other conditions.

The Probability of a Global Market Entering a Bear Market is Relatively Low

Since the 1970s, there have been five instances of wars causing significant oil price fluctuations: the first oil crisis in 1973, the second oil crisis in 1978, the Gulf War in 1990, the Kosovo War and coordinated production cuts in 1999, and the Russia-Ukraine conflict in 2022. A closer examination of these five events reveals that each unfolded differently and was influenced by numerous unique factors.

The first two oil crises in 1973 and 1978 had similarities. Before these crises, US listed company earnings, economic fundamentals, and inflation were already significantly rising, with inflation further escalating due to additional factors. In other words, before the outbreak of war and the surge in oil prices, the US economy was already overheating, the Federal Reserve had begun raising interest rates to curb inflation, and the economic cycle was in its mid-to-high phase. At this point, the oil price surge became the straw that broke the camel's back.

The third event was different. Before the Gulf War broke out in August 1990, the US economy was already in a downturn, and the Federal Reserve had begun cutting interest rates, with the yield curve already trending downwards. (According to the official recession periods identified by the National Bureau of Economic Research of the United States), the US had already entered a recessionary period before the 1990 Gulf War. The Gulf War led to a surge and subsequent fall in oil prices, and by the time oil prices returned to their original levels, the economic recession had largely concluded. The characteristic of the third event was that the US was already in recession before the war began.

The fourth event was the Kosovo War in 1999, which was a highly unusual occurrence. Its uniqueness lay in the implicit presence of a major event – the dot-com cycle, which led to structural explosive growth in the US economy and corporate earnings. In 1999, oil prices surged from $10 to $30, a threefold increase; simultaneously, the Federal Reserve raised interest rates six consecutive times. During this period, the structural explosive growth of the US economy was not interrupted, with corporate earnings growth corresponding to this structural boom exceeding 60%. Consequently, the Nasdaq index rose by 90% against the backdrop of six consecutive rate hikes.

The last instance was the Russia-Ukraine conflict in 2022. Following the outbreak of war, the Federal Reserve began a series of 11 consecutive rate hikes at an unprecedented pace starting in April. 2022 and 2023 were periods of strongest market expectation for a US economic recession. However, the US economy withstood the pressure throughout these 11 consecutive rate hikes. This resilience can be partly attributed to significant expansion of US fiscal policy (including the three major legislative acts), which successfully counteracted the impact of consecutive rate hikes and high oil prices.

By reviewing these five historical experiences, we find that the logical transmission path from war, oil price increase, to economic recession, and subsequently affecting US corporate earnings and fundamentals, involves numerous factors that require consideration. For example, what was the position of the US economic cycle before the war? Was there room for maneuver in inflation and monetary policy? Did the US have structural economic growth drivers? Would fiscal policy provide significant countercyclical support?

All these factors play a crucial role. Returning to the latest situation, I am relatively optimistic. Currently, there are many positive factors that lead us to judge the probability of a US economic recession as low.

The first positive factor is structural economic expansion, similar to the tech and internet cycle of 1999, driven this time by artificial intelligence, including improvements in total factor productivity. The second positive factor is political expediency. In 2022, the Biden administration introduced extremely strong fiscal stimulus to prevent a recession before the general election. The current situation is similar, with electoral pressures potentially prompting the US to continue with larger-scale fiscal expansion.

The third point is even more significant. The characteristic of the US economy since 2020, distinguishing it most from previous cycles, is that it does not experience strong upswings leading to recovery, nor sharp downturns leading to recession, a stark contrast to the cyclical boom-and-bust patterns of the past. In simpler terms, there are no sharp rises without sharp falls.

Over the past four years, as we have analyzed the structure of US corporate earnings and economic fundamentals, we have observed a continuous rotation of support points. Sometimes it came from government infrastructure investment, sometimes from consumer services, sometimes from real estate, and at other times from investment in goods or artificial intelligence. The support for earnings has always been provided by different variables, not a simultaneous broad-based explosion across all directions within the same period, thus leaving room for maneuver and flexibility in the economy.

Fundamental Trends Are Always the Most Important

The second judgment concerns liquidity and interest rates. Concerns exist about how sustained high oil prices might affect US liquidity and expectations for interest rate adjustments. Let me revisit the situation around the 1999 tech and internet cycle for a comparative analysis.

From 1999 to 2000, oil production cuts combined with the Kosovo War caused oil prices to rise from $10 to $30. Concurrently, the US implemented six consecutive high-interest rate hikes, from 4.75% to 6.5%.

How did US assets perform during this period? Firstly, after the Federal Reserve's first rate hike in June 1999, the Dow Jones Industrial Average, representing broad demand in the traditional economy, entered a period of consolidation, suppressed by high interest rates, high oil prices, and high inflation. On the other hand, the Nasdaq index, representing structural economic explosive growth, experienced a brief adjustment after the rate hikes but subsequently surged by 90%. This was attributed to its sufficiently rapid earnings growth.

Reviewing the earnings per share (EPS) growth of the Nasdaq 100 index throughout the 1990s tech and internet cycle reveals a consistent and healthy EPS growth rate of 20% to 30% during the widespread adoption of computers, software, and Windows 95. However, in 1999, there was an explosive surge, with the Nasdaq 100 index experiencing earnings growth of up to 60%, and many individual stocks doubling their earnings.

Underneath such rapid localized explosive growth, an counter-intuitive phenomenon occurred: the market actually awarded higher valuation multiples. Its trailing P/E ratio reached 95 times, and its forward P/E ratio reached 65 times, both the highest points of the 1990s. Usually, interest rate hikes and liquidity contraction severely suppress valuations, but the ultimate outcome was that funds formed concentrated positions. When the market cannot find other sectors with high prosperity, localized high prosperity becomes the primary direction for capital concentration.

This situation has also occurred in A-shares. In 2013, the domestic macroeconomic environment was characterized by a "money crunch," with short-term interest rates soaring above 4.5% and liquidity extremely tight. Yet, the ChiNext index experienced a primary upward trend throughout 2013, with gains approaching 80%. The reason behind this was the explosive growth in mobile game revenue and M&A activities at the time, which supported earnings growth similar to the Nasdaq in 1999, thereby leading to structural capital concentration.

The core of these cases lies in the fact that when fundamental trends diverge from liquidity and interest rate trends, fundamental trends are always the most important. Only with sufficiently rapid growth and growth expectations can capital be attracted for concentration.

Looking back at the 1999 tech and internet bubble, its eventual burst was not due to interest rate hikes, but because after the financial report disclosures in April 2000, the market discovered that capital expenditures and earnings were not continuing to grow. In hindsight, the earnings growth of the Nasdaq 100 index in 2000 was only 12%. Therefore, its rise relied on earnings explosions and expectations, while its collapse stemmed from fundamentals falling short of expectations and the invalidation of the capital expenditure thesis, having little to do with the interest rate cycle.

Structural Allocation in the Second Quarter is Clearly More Significant Than Overall Positions

Based on this, looking ahead to the second quarter and the second half of the year, we can summarize the following conclusions.

First, entering the second quarter, the most important strategy is structure; structural allocation is clearly more significant than overall positions. Data shows that starting from the second quarter each year, the dispersion of industry performance rapidly expands, continuing through May to July. The expansion of dispersion implies that inter-industry performance differentiation is extremely large.

Currently, as long as two points are confirmed – that the global economy will not enter a recession in the second half of the year and the AI trend will not collapse – then during the "April Decision" period, it is feasible to identify sectors with structural high prosperity. Of course, if one pessimistically believes that the global economy will inevitably decline and enter a bear market, then there is no need for an "April Decision"; instead, one should continuously reduce positions.

For the "April Decision" allocation direction, we should look for industries that are less affected by high oil prices and high interest rates, possess independent potential for localized explosive growth, and can pass on cost pressures downstream. The main directions are as follows:

First is new energy. The fundamentals of new energy have gradually improved before the Iranian issue erupted. The most typical representative is residential energy storage, specifically the export of inverter equipment. Before October last year, its export growth was in the single digits. However, with the introduction of subsidy policies by major economies in Europe and Australia in the fourth quarter, the inflection point for export growth occurred in November and December, reaching over 20%. Furthermore, domestic large-scale energy storage is expected to grow very rapidly this year, reaching about 70%, and will maintain good growth next year. Therefore, the first main theme is energy storage (corresponding to overseas residential storage and domestic large-scale storage). This sector is currently at the bottom of a three-year cycle, just showing an upward inflection point. Coupled with the megatrend of energy security, new energy storage is an important allocation chain.

Second is the explosion of domestic AI data centers, and the outbreak of domestic AI applications and token consumption volume. The latest data has been updated to March. Last week, Volcengine held its first launch event of the year, updating various measurement data. Taking ByteDance's Doubao large model as an example, its daily average token (yuan) calls in its mobile application were 30 trillion in September last year (2025); by December 2025, the daily average calls reached 60 trillion, doubling; and by March 2026, the daily average calls increased to 120 trillion, doubling again. On a high base, it has achieved continuous doubling growth every quarter. Moreover, this is before the implementation of video multimodal models like Seedance 2.0. In the future, as multimodal large models further increase token consumption, their growth slope may become even steeper.

In the past two years, China's AI industry chain has mainly been driven by thematic investment. However, this year is likely to be a breakout phase with a steeper growth slope for token consumption, marking the "Year One" of agents and the widespread deployment of data centers in China. They are likely to transition from a state of high apparent valuations and primarily thematic investment to a substantive state driven by industry trends and an explosion in orders this year. From the perspective of application and capital expenditure growth, China is beginning to catch up with overseas markets this year. Overseas cloud service giants' capital expenditure growth in 2024 has reached 60% to 70%. We have been relatively lagging in the past two years, but our catching-up speed this year will be very rapid. China's AI industry chain (including data centers and corresponding domestic alternatives for chips and semiconductors) is experiencing explosive order growth, but current apparent valuations are high, requiring stabilization of market risk appetite. Current risk appetite has indeed stabilized at the bottom, providing significant support for this.

The third direction is overseas computing power. The market has some concerns here: can the high growth rate of capital expenditure continue? The source of concern lies in whether the cash flow and financing costs of major companies can continue to support high capital expenditures. In the past few years, the capital expenditures of tech giants have been partly driven by FOMO (fear of missing out) – the fear of being left behind if they don't invest. If this anxiety is the sole driver, capital expenditure will become unsustainable once cash flow is depleted or financing costs become too high. However, since the beginning of this year, the growth slope of global token consumption has seen a rapid surge. According to partial data from third-party platforms, from March 2025 to the present, especially after the integration of more agents and enhanced coding capabilities in February 2026, the growth slope of token consumption has become incomparable to before. If subsequent financial reports show that the profitability of overseas cloud giants improves due to the surge in token demand and price increases, then capital expenditure will no longer be driven by "fear of missing out" but by profit improvement. At that point, the cash flow situation and the level of financing costs will become less important.

Hong Kong Stocks Have an Opportunity for Valuation Repair

Finally, let's spend a few minutes looking at Hong Kong stocks. Hong Kong stocks are in a very special situation. We previously mentioned that East Asian capital markets, including China's A-shares, Taiwan's stock market, Japan's stock market, and South Korea's stock market, all have a large number of industries directly linked to overseas hard technologies. Whether it's optical modules, printed circuit boards (PCBs), optical chips, multilayer ceramic capacitors (MLCCs), semiconductor materials, or storage, their earnings have been explosive.

However, the Hong Kong stock market lacks these overseas hardware and hard technology components. Taking the well-known Hang Seng Tech Index as an example, its profit drivers are mainly e-commerce, food delivery, gaming, advertising, and automobiles. This essentially represents the broadest demand in China. Currently, this broad demand is performing moderately, as China is one of the few among the world's top 20 economies that has not undertaken fiscal expansion, maintaining a relatively conservative fiscal policy to leave more room and flexibility for the future.

Therefore, against the backdrop of no significant fiscal expansion and moderate broad demand, the Hong Kong stock market faces the challenge that its profit elasticity is the weakest among the five East Asian markets.

In fact, before the conflict in Iran occurred this year, Hong Kong stocks had already weakened. The reason is that Hong Kong stock earnings reporting periods differ from A-shares; major financial reports are released in March. Investors were concerned about poor profit performance and lack of elasticity, leading to no buying interest in Hong Kong stocks in January and February, causing an early downturn.

However, the current positive aspect is that after half a year of continuous decline, market expectations for poor performance in the March earnings season have been largely digested. The second quarter presents a rare window of respite because no financial reports are released during this period. Additionally, given that Hong Kong stocks have already fallen for half a year, if global market risk appetite stabilizes, Hong Kong stocks will have an opportunity for valuation repair in the second quarter.

However, for Hong Kong stocks to experience a larger-scale upward trend in the medium term, or to reduce volatility and increase the Sharpe ratio, investors looking to hold indices like the Hang Seng Tech will still need to wait for the emergence of profit elasticity, which means waiting for the recovery of broad Chinese demand. Before this happens, the entire Hong Kong stock market will only be experiencing valuation fluctuations, leading to a relatively poor holding experience, making it suitable only for bottom-fishing in the second quarter. Once it rises, profit-taking will be necessary. However, the second quarter conveniently provides such a feasible window.

Above are our updated judgments on the short-to-medium and long-term perspectives.

To briefly summarize the core conclusions, I remain relatively optimistic about the overall market trend. For equity assets, the core judgment assumption here is that the major industry trend of artificial intelligence (AI) has not ended. This is the core soul of East Asian markets (including A-shares). The core logic that this industry trend has not ended lies in its own industry, including tokens, applications, and agents, all experiencing an outbreak.

Secondly, we do not believe that the US will experience a substantial recession in the second half of the year under the current circumstances; its economic resilience remains strong. Against this backdrop, it is sufficient to conclude that the global market will not enter a major bear market. Therefore, our core task in the second quarter is to identify directions of structural high growth. The most significant characteristic of the second quarter is "structure far outweighs positions"; the severity of buying the right or wrong structure is far greater than the level of positions held.

Looking ahead to the second quarter and the next six months, we have screened several sectors where growth, profits, and orders are expected to explode. For example, new energy related to energy storage, overseas computing power, and domestic industry chains such as AI data centers (AIDC) and domestic alternatives for semiconductors.

Risk disclosures and disclaimers

Markets are risky, and investments require caution. This article does not constitute personal investment advice, nor does it consider the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are appropriate for their specific circumstances. Investment based on this is at your own risk.