Where to Begin

On July 27th, during a closed-door meeting held on Monday, Morgan Stanley China's Chief Economist, Xing Ziqiang, shared his latest insights on topics such as whether the "September 24th moment" would reoccur and the progress of the AI industry. Xing Ziqiang assessed that the current policy environment does not yet resemble the critical turning point seen in September 2024.

Although the economy's growth in the second quarter slowed to around 4.3%, exports—especially new "three new items" like AI hardware—are providing a cushion for the economy. Over the past six months, AI-related hardware infrastructure has contributed more than 10 percentage points to China's export growth. He anticipates that the high-level meeting this week will focus on fine-tuning and accelerating the implementation of existing policies.

Meanwhile, U.S. AI investments are hitting a "double wall" of political and physical obstacles. Factors such as local community opposition, electricity shortages, and legislative hurdles are creating strong resistance to data center construction in the U.S. Last year, $156 billion worth of data center projects were canceled or postponed, and in just the first quarter of this year, another $130 billion in projects have stalled or been completely abandoned.

In contrast, China's AI sector is seeking alternative ways to compensate for computing power shortages, adopting a systematic approach to break through. Combined with its inherent advantages in physical AI and embodied intelligence, China is carving out a differentiated path.

Additionally, Chief Strategist Laura Wang noted that since July, the Hang Seng Index has outperformed the A-share CSI 300 Index by nearly 15 percentage points. In the current market environment, she still recommends short-term bullish positions on Hong Kong stocks. Furthermore, the recent listing of Changxin Memory Technologies is beneficial for the A-share market's ecosystem. While it may have a short-term absorption effect on market liquidity, it remains positive for the long term, as the current AI super-cycle is far from over. Overall, the A-share market is currently lukewarm, and it is advisable to exercise caution regarding short-term liquidity fluctuations. A better opportunity for a large-scale market entry may emerge as autumn approaches.

AI Hardware Undergoes Volatile Transformations

El Niño Challenges Central Bank Policy and Liquidity

This summer has been highly turbulent, with global capital markets filled with anxiety and divergence. This week, we will continue to share and delve deeper into analyses, including whether the highly anticipated domestic economic work meeting will create a strong turning point similar to September 2024, boosting market confidence and economic support. We have also observed that major global sectors, including U.S. stocks and AI hardware-focused markets in Taiwan, China, and South Korea, have experienced a significant shakeout over the past month and a half.

The World Artificial Intelligence Conference (WAIC) in Shanghai has just concluded, and our analysts conducted thorough on-site research, bringing back new findings. At the same time, geopolitical tensions in the Middle East and energy shocks seem to be resurging. Moreover, as we have analyzed over the past month, a special phenomenon—the confirmed return of a super El Niño—is occurring this year and next, which will severely test global inflation, central bank policies, and liquidity shifts. Today, we will analyze these topics together.

The Sept 24 Moment Reappears?

Policy May Be Hard to Exceed Expectations

First, the question of whether the "September 24th moment" will reoccur arises. Economic growth in the second quarter slowed to around 4.3%, below the annual target. Many investors are closely watching the high-level meeting scheduled for this week. Our assessment is that the current situation is not yet a critical turning point like September 2024. The meeting this week is likely to focus on fine-tuning and accelerating the implementation of existing policies, rather than being a strong inflection point like September 24th.

At the core of this is exports, which have been very strong recently due to global AI capital expenditures and energy transition investments. These two super-cycles provide robust dual support for Chinese exports. For example, over the past six months, AI-related hardware infrastructure has contributed over 10 percentage points to China's export growth, and the "new three items" in energy—new energy vehicles, lithium batteries, and energy storage equipment—have contributed similarly. Together, these provide a certain cushion for current economic growth. The focus remains on utilizing the remaining fiscal policy space for this year, determining how much quota is available and which sectors, particularly the "six networks," can generate short-term workloads.

U.S. Data Center Construction Encounters Political and Physical "Double Wall"

Returning to the global volatility mentioned earlier, the narrative has shifted. With the sharp correction in AI hardware sectors, the market is also questioning whether the AI narrative has changed. Over the past month, we have analyzed multiple times that there are structural divergences in the AI industry between China and the U.S. On the U.S. side, while internet tech companies still have ambitious AI capital expenditure plans—either raising or maintaining spending in recent earnings reports—there are challenges in actually implementing these expenditures into workloads.

The U.S. is encountering both political and physical "walls." For instance, despite these cloud companies collectively planning to invest $870 billion in data centers this year, data center construction in the U.S. is facing strong political resistance. These include three "P"s: public opposition from local communities concerned about rising electricity bills, water resource and environmental pressures, and noise from large construction projects. More critically, most Americans believe these data centers do not create local jobs, leading to opposition. The second "P" is power; the U.S. is severely short on electricity. Without sufficient power, and due to community opposition, many data centers are forced to disconnect from the public power grid, causing further delays. Finally, "politics": several states have issued temporary moratoriums banning data center construction. Last year, $156 billion in data center projects were canceled or postponed. In just the first quarter of this year, nearly $130 billion in projects have stalled or been completely abandoned. This is a classic case of supply-side risks falling short of expectations.

Domestic AI Takes a Different Path

China's situation is different. As we have analyzed before, China's open-source large language models represent a different path. The recent WAIC in Shanghai also showcased a clear dividing line. China's AI is attempting to compensate for computing power shortages through alternative methods, adopting a systematic breakthrough. First, Chinese AI competition is moving upward, shifting from single-chip specifications to super-large cluster architectures like SuperPoD. To compensate for the shortcomings of single-chip processing power, Chinese AI hardware manufacturers are leveraging system-level networking and rack design advantages. This was a major takeaway from WAIC last week—the systemic expansion of computing hardware. In areas where China already has an advantage, such as physical AI and embodied intelligence, the industry is accelerating its deployment. Robotics and intelligent hardware were crowded tracks at WAIC, accounting for nearly a quarter of exhibitors. Industrial AI agent solutions are expanding into engineering automation, and their profit models are improving. The core bottleneck of data is also receiving attention and solutions in China.

El Niño's Inflation Risk Marginally Affects AI Hardware Trading

We also note that the global AI narrative is being influenced not only by the aforementioned constraints in the U.S. but also by the resurgence of geopolitical tensions, particularly the ongoing conflict in the Middle East, combined with extreme weather events—the strong return of El Niño. The return of El Niño has clear implications for asset allocation. Currently, heatwaves in Europe, persistent heavy rains in southern regions, and an 81% probability of a very strong El Niño event continuing in the second half of the year all pose risks. The inflation risk, especially for food and commodities, is significant. Monetary policy is under pressure, and the question of whether to raise real interest rates in the face of food inflation arises. These factors marginally affect the already crowded AI hardware trading, particularly since AI companies are heavily financing in credit, equity, and capital markets. Given the massive scale of financing—potentially trillions of dollars annually—these companies are highly sensitive to marginal changes in liquidity, including the impact of El Niño on global interest rates, liquidity, and inflation.

Morgan Stanley China Chief Strategist Laura Wang: Recommends Short-Term Bullish on Hong Kong Stocks

Hong Kong Stock Q2 Earnings May Exceed Expectations Overall

Several weeks ago, we began recommending a shift in focus within Chinese stock allocation from A-shares to Hong Kong stocks. Since July, the Hang Seng Index has outperformed the A-share CSI 300 Index by nearly 15 percentage points. Overall, we have captured this inflection point well. In the current market environment, we maintain a short-term bullish stance on Hong Kong stocks while avoiding A-shares to a certain extent. However, for the long term, we believe the structural advantages of the A-share market's ecosystem are strong. Short-term momentum in Hong Kong stocks is supported by several factors. The pressure from July's IPO lock-up expirations has been fully released, and overall, we expect Q2 earnings to exceed market expectations.

Last week, we published several reports related to today's discussion. One analyzed the earnings pre-announcements from listed companies in Q2, which we have collected. We saw improvements in both A-shares and Hong Kong stocks. However, the change in A-shares largely involved a reduction in the scale of negative earnings warnings. In contrast, MSCI China, which focuses more on offshore-listed Chinese companies, showed a clear positive trend, with nearly 14 percentage points of positive earnings pre-announcements. Therefore, we believe the expected improvement in Hong Kong stocks' Q2 earnings compared to Q1 and the second half of last year is likely to be validated.

Furthermore, global AI investment opportunities are shifting from the front end toward application and deployment. In this process, we see that Chinese AI giants listed in Hong Kong have significant opportunities in commercialization, ecosystem establishment, and future expansion. Thus, we still recommend a more concentrated allocation to Hong Kong stocks in the short term. Regarding A-shares, we also published a report on the A-share market sentiment index last week. The current trading volume and sentiment in the A-share market are lukewarm and not very heated. This is due to two reasons: one is a short-term policy vacuum. In such a policy vacuum, it is reasonable for the market to adopt a wait-and-see approach and stay on the sidelines.

Changxin Memory Listing Benefits A-Share Ecosystem

Autumn May Bring a Large-Scale Entry Opportunity

Additionally, on July 27th, the ultra-large Changxin Memory Technologies began trading. We believe this is a significant positive for the A-share ecosystem in the long term, but in the short term, it does have a certain absorption effect on market liquidity. We provided warnings about this in previous weeks. This absorption effect may persist for a short period, such as one to two weeks. We are bullish on it in the long term because we believe the current AI super-cycle is far from over. However, from the start of the year to now, despite substantial gains in related A-share sectors and stocks, the proportion of individual retail investors benefiting from these gains is relatively low. This is because investing in the AI super-cycle requires a relatively high threshold of related professional knowledge and overall sector allocation understanding. So far, retail investor participation is limited. Against this backdrop, the listing and trading of Changxin Memory is an excellent event for long-term education of investors on the sector and main investment themes. Therefore, in the long term, we view Changxin Memory and similar ultra-large, high-quality Chinese companies as positive for attracting foreign capital. However, in the short term, it is still necessary to avoid liquidity-related risks. Hence, we maintain our recommendation to focus overall assets on Hong Kong-listed internet giants and related sectors for a period of one to two months. In the A-share market, we hope to avoid short-term liquidity fluctuations and volatility from the global AI super-cycle. As we approach autumn, there may be a better opportunity for a large-scale market entry.