Recently, Chen Guo, Deputy Head of the Research Institute and Chief Strategist at East Money Securities, shared his outlook on the A-share market's performance in the second half of the year during a program. The key points are summarized as follows.
1. (Overseas Market Volatility) For China's stock market, I believe this does not constitute a systemic impact. Sectors highly correlated with U.S. AI stocks might be affected, but for most stocks, it's more about sentiment transmission. Today, most A-shares are adjusting, but this phenomenon will not last long. Initially, sentiment transmission leads to broad-based adjustments or declines, but the market will gradually stabilize and see differentiation. If there are new positive developments in U.S. AI, related sectors will rebound significantly; even without them, other sectors will stabilize quickly and then develop differentiated trends based on their own fundamentals.
2. Many A-share stocks performed weakly or even declined significantly in the first half of this year. Some of this relates to fundamentals, but a considerable portion, or the vast majority, has little to do with fundamentals. I believe there has been unjustified selling, and a recovery is expected subsequently.
3. In the second half of this year, especially the third quarter, economic expectations and capital inflows are both showing marginal improvement. Downside space is limited, while continued marginal improvement provides upside potential. Of course, overseas disturbances will occur during the process, potentially forming a strong, volatile trend with an upward trajectory. The market structure may also differ from the second quarter. The second quarter, from March to May, saw concentration in the silicon-based upstream sector. June saw a slight rebalancing, but it's still in an initial, hesitant state. Simply put, the market structure will further rebalance in the third quarter.
4. From March, April, May to early June, market expectations for asset pricing were too extreme, showing a K-shaped divergence—expectations for the silicon-based economy, especially the U.S. AI capex chain, were crowded, possibly high, making further positive surprises difficult, with potential for downward revisions. On the other end, expectations for China's carbon-based economy were also extrapolated, with overly pessimistic pricing. Subsequent data will lead to corrections; both lines will see revisions in fundamentals and expectations in the second half, leading to rebalancing. At least, the three directions of U.S. AI, China AI, and China's carbon-based economy need rebalancing. U.S. AI-related directions may be less optimistic, China AI is okay, slightly better than market expectations, and China's carbon-based expectations should be revised upward. If these factors are confirmed in the second half, the market will consider rebalancing.
5. The outlook for the carbon-based economy is not so pessimistic, while silicon-based, especially U.S. silicon-based, is not as optimistic as imagined. However, China's AI follows a cost-effective path, with higher probability of commercialization success and return on investment because we haven't over-invested. Large models like DeepSeek are competitive with overseas advanced levels, offering much better value for money, so confidence in China's AI is stronger.
Chen Guo also shared this view in his latest research report, stating that while short-term external volatility continues, affecting A-shares with adjustments, the main external changes are rising concerns about the U.S. AI industry chain and expectations of interest rate hikes and liquidity tightening. Although A-share valuation environment may be disturbed short-term, overall, there is no systemic risk. In the third quarter, the upside potential for A-share indices is greater than the downside risk.
Chen Guo pointed out that short-term internal structure in A-shares is crowded, with valuation divergence at historically high levels. This may stem from overly pessimistic expectations for the carbon-based economy or traditional industries, which are expected to improve marginally in the third quarter, so gradual structural rebalancing can be considered.
Which specific areas to focus on next?
Chen Guo noted that recently many traditional industry companies have been intensively increasing share buybacks and repurchases. Historically, announcements of buyback plans have shown some correlation with stock price bottoms. He believes such companies could also be considered for rebalancing in the next stage.
Key focus industries: essential consumption (food, etc.), healthcare (CXO, etc.), coal, power equipment & new energy, non-bank financials, new consumption, internet, semiconductor equipment/materials/domestic AI chain, etc.
Thematic directions: focus on blue-chip combinations with active buybacks and share increases.
Today, most A-shares are adjusting, but it will not last long.
Host: First, a topic of great concern to viewers—today the market saw some adjustment. How do you view the current pullback? Is it sustainable?
Chen Guo: I think it may also be caused by overseas market fluctuations. The global stock market's focus is AI, and the AI industry chain has certain linkages. Recently, there have been some expectation fluctuations in the U.S. AI chain. For example, OpenAI originally planned an IPO this year, with the market expecting impressive data before the IPO to secure financing for faster AI capex investment. However, reports suggest OpenAI is considering canceling this year's IPO plan, postponing it to next year. This may affect expectations about its revenue growth—temporarily not advancing the IPO, though other methods will be used, it's hard to lock in massive financing like an IPO, thus affecting subsequent investment pace. This impacts U.S. stocks and also affects global AI-related sectors.
Second, Apple's products have seen significant price increases, driven by silicon-based upstream inflation from AI. But the market worries whether such cost-driven passive price hikes can be fully passed on to consumers—or whether the strong inflation in silicon-based upstream is sufficiently balanced with silicon-based downstream and carbon-based economies. The market is concerned that the international economy and downstream sectors cannot bear international upstream inflation, which also constitutes a volatility factor.
Additionally, on the denominator side, concerns arise about whether the Fed will initiate rate hikes and how many hikes. Overall, concerns are mainly overseas-related, especially regarding the numerator side of the U.S. AI industry and the denominator side of Fed rate hike decisions.
For China's stock market, I believe this does not constitute a systemic impact. Sectors highly correlated with U.S. AI stocks might be affected, but for most stocks, it's more about sentiment transmission. Today, most A-shares are adjusting, but this phenomenon will not last long. Initially, sentiment transmission leads to broad-based adjustments or declines, but the market will gradually stabilize and see differentiation. If there are new positive developments in U.S. AI, related sectors will rebound significantly; even without them, other sectors will stabilize quickly and then develop differentiated trends based on their own fundamentals.
Many A-share stocks were unjustifiably sold in the first half, with recovery expected subsequently.
Host: Although the market has pulled back short-term, you recently stated strategically not to be bearish on A-shares, with upside potential far exceeding downside in the next quarter. Do you judge the current A-share adjustment as a rare window for positioning? What are the core underlying logic and key variables supporting potential continued strength in A-shares?
Chen Guo: Many A-share stocks performed weakly or even declined significantly in the first half of this year. Some of this relates to fundamentals, but a considerable portion, or the vast majority, has little to do with fundamentals. I believe there has been unjustified selling, and a recovery is expected subsequently.
Recovery may have two catalysts: first, the market is overly pessimistic about the carbon-based economy, such as negative growth in May's social financing and retail sales, with simple trend extrapolation. Actually, May data had special factors—last May's retail sales growth was over 6%, an extremely high base. A single month's negative growth doesn't represent the overall real consumption situation. Our models show that June, July, and August will see continuous marginal improvement, with carbon-based economy supported and improving marginally.
Second is the capital side. In the first half, overall A-share capital tended to outflow, with many broad-based indices still in a new bull market. But looking at weekly sequential data, net outflows are weakening. For directions related to broad-based indices or blue-chip white horses, the market may think previous adjustments were overdone, showing some value, with value investment funds slowly positioning. Many listed companies (mostly traditional carbon-based industries) have major shareholders and executives starting to increase holdings and buybacks, also bringing net inflows and boosting other funds' confidence in these companies' value investment or allocation basis. A net inflow cycle may gradually start, at least structurally, reflected in some traditional blue-chip white horse sectors that performed weakly in the first half.
Downside space limited in Q3, more upside potential, structure further rebalancing.
So I believe in the second half of this year, especially the third quarter, economic expectations and capital inflows are both showing marginal improvement. Downside space is limited, while continued marginal improvement provides upside potential. Of course, overseas disturbances will occur during the process, potentially forming a strong, volatile trend with an upward trajectory. The market structure may also differ from the second quarter. The second quarter, from March to May, saw concentration in the silicon-based upstream sector. June saw a slight rebalancing, but it's still in an initial, hesitant state. This is because people also hesitate whether capital is存量or减量, related to whether fundamental divergence intensifies. According to our judgment, both are improving marginally, and structure will also rebalance. Simply put, downside space is limited in the third quarter, upside potential is greater, and structure will further rebalance.
Carbon-based not so pessimistic, silicon-based not so optimistic.
Host: You just mentioned rebalancing, and recently A-shares have indeed seen style rotation. Your latest view also emphasizes this round of market movement will move toward rebalancing, no longer with a single AI sector dominating. What do you think is the core driving force behind this round of style rotation?
Chen Guo: I think from March, April, May to early June, market expectations for asset pricing were too extreme, showing a K-shaped divergence—expectations for the silicon-based economy, especially the U.S. AI capex chain, were crowded, possibly high, making further positive surprises difficult, with potential for downward revisions. On the other end, expectations for China's carbon-based economy were also extrapolated, with overly pessimistic pricing. Subsequent data will lead to corrections; both lines will see revisions in fundamentals and expectations in the second half, leading to rebalancing.
Specifically, the issue with the U.S. industry chain is whether investment returns can be recouped. U.S. IT investment as a percentage of GDP has exceeded the 2000 peak. Of course, exceeding doesn't necessarily mean a top, but 2000's IT investment share was also a historical high, and it proved difficult to recoup short-term returns afterwards, with investment透支, and当时over-optimism about the internet economy. Long-term internet space is large, but short-term obtaining widespread profits was still difficult. Now it's similar; AI is not completely the same as the internet, but upstream capital expenditure is large, IT investment share even higher than the 2000 peak, with doubts about whether returns can be recouped, and current returns are not as good as the internet era—in 2000, a batch of internet companies could profit, look at EPS, like Yahoo, eBay, but overall investment return rate was not high, considered过度超前. The current problem is, major U.S. AI companies are mainly two new forces, Anthropic and OpenAI. Google etc. have related business in platforms, but large models themselves haven't generated positive free cash flow and profit. Musk's Grok has been merged into SpaceX, basically半放弃状态. AI profit situation is not ideal, and Anthropic guides inability to profit in the second half; Q2 single-quarter profit was due to Musk's low-cost computing power lease优惠; OpenAI also stated inability to profit, even IPO financing delayed, short-term reporting strong revenue growth data is difficult, recently even talking about price wars. So U.S. AI economy may not exceed expectations, previous expectations may have been too optimistic. Subsequently need to monitor whether downstream manufacturers' revenue expectations are revised down, and if so, whether it affects capital expenditure expectations—investing so much ultimately wants to recoup returns; if making money harder, future investment may marginally decrease.
Conversely, China's carbon-based economy is similar; people only see some同步indicators like May's negative retail sales growth, but the economy has leading and lagging indicators, it's a cycle. The core of China's domestic demand carbon-based economy is asset-side repair, including whether housing price decline速度slows, whether some cities stop falling or even rebound. The past half-year situation is improving, first-tier cities have stopped falling and rebounded, strong second-tier也开始出现, most second-tier cities' decline趋缓, which is crucial for走出资产负债表衰退. First and second-tier price stability, already accounting for a considerable proportion of household balance sheets, can drive overall consumption. Another aspect is income expectation; the market过度高估carbon-based and silicon-based opposition, thinking AI development will lead to碳基income concerns. But China's AI model differs from the U.S.; U.S. AI确实will冲击替代, China更多treats AI as a tool, and social system differs, won't cause large-scale unemployment,更多using silicon-based to empower carbon-based. The market may过度apply U.S. situation to China. Additionally, PPI, industrial enterprise profits, company profit growth等leading indicators this year are better than last year, will slowly传导to income expectations and consumption. Consumption is a lagging indicator, improving only after asset prices and enterprise income improve; the stock market is leading, should look at leading indicators. So carbon-based not so pessimistic, silicon-based especially U.S. silicon-based not as optimistic as imagined. But China's AI follows a cost-effective path, with higher probability of commercialization success and return on investment because we haven't over-invested. Large models like DeepSeek are competitive with overseas advanced levels, offering much better value for money, so confidence in China's AI is stronger. Conversely, China's AI progress also makes the future of U.S. AI's ability to maintain high prices, high income, recoup high investment more questionable, the rationality of high capital expenditure growth more doubtful. Therefore, investment positioning needs to reconsider balance. At least, the three directions of U.S. AI, China AI, and China's carbon-based economy need rebalancing. U.S. AI-related directions may be less optimistic, China AI is okay, slightly better than market expectations, and China's carbon-based expectations should be revised upward. If these factors are confirmed in the second half, the market will consider rebalancing.
These directions will show performance in the second half, such as new and old energy, non-bank financials, etc.
Host: Overall, the market rebalancing logic is sufficient, with many driving forces. But from market performance, recently A-shares are moving from extreme divergence toward rebalancing, but the switching process has been反复. What do you think the market is纠结about regarding rebalancing or style rotation currently?
Chen Guo: We have extensive交流with institutional investors, with two representative divergences. One is divergence over whether U.S. AI can exceed expectations again. In the first half, U.S. AI indeed exceeded expectations; Anthropic's Cloud or AI agent revenue growth exceeded expectations, driving capital expenditure further upward revisions, various IT hardware景气strong, supply-demand tight, U.S. AI capex growth about 80%-100%, a high景气sector. People don't rule out growth may go up again in the second half, but precisely judging景气度changes is difficult. My view is it's challenging—this year Anthropic found the programmer AI agent替代sector, but next areas like finance, law, high-salary positions or video editing等aren't that easy, it hasn't given clear timetable and路线图, may complete this milestone financing before thinking next step, growth not that fast. OpenAI also next year, if financing delayed, big moves可能next year出, short-term can't出, so存在judgment差异. Second, China's carbon-based also has different judgments. Many think China's economy still has challenges, like housing price decline continuing, population decrease pressure, income growth困难, consumption confidence and场景supply未必significantly improve, therefore switching logic insufficient. My view is, consumption improvement肯定slow, currently景气at low levels, but关键是marginal improvement and pricing匹配—景气low到不能再低and marginal improvement, also has investment value. Middle ground like new and old energy and non-bank financials, we重点推荐, recognition稍高but not坚决or未形成consensus. Core reason is first-half持股体验不好—not switching, staying in AI especially U.S. AI capex chain赚钱效应good, people趋势外推,舍不得look other directions. New and old energy affected by霍尔木兹海峡oil price fluctuations,交易扰动多, first-half业绩不错but资金流出,股价没涨, people信心and耐心不足. But first-half factors不能推演到second half, these扰动second half不太存在. Therefore, new and old energy like coal, lithium battery, energy storage等growth不错, mid-year reports many重点companies有50%-100% growth, securities companies growth also good, investors普遍认可but担心growth好能否上涨,信心不足. But站在value investment, profit valuation匹配angle,结合second half environment与first half不同, we believe these directions will show performance in the second half. Rebalancing or switching process,资金will陆续尝试,走势变化后更多资金follow,起初slow or曲折,往后循环will走强.关键是fundamentals, valuation marginal improvement是否成立, if成立,表现probability较大. Recently虽有rebalancing迹象but纠结, this is初期natural phenomenon.
Two paths for China-U.S. AI: carbon-based benefit and carbon-based substitution.
Host: Next, please深入分析silicon-based and carbon-based. You previously明确指出don't treat AI as everything,提醒investors facing AI牛市不能sleep, and pointed out U.S.部分AI giants' IPO闻到bubble味道.同时强调now won't short AI, especially A-share technology具备独立行情cycle. Can you通俗解读core differences between Chinese and foreign AI for viewers? What are the supporting壁垒for A-shares' independent行情?
Chen Guo: The core of China's AI is pursuing碳基从中获益, technology以人为本, silicon-based empowering carbon-based, ultimately silicon-based not高成本operation, electricity, computing power, models not expensive business models,更多mass广泛使用. U.S. emphasizes AI agent,毫不掩饰used for替代carbon-based labor, will bring large-scale unemployment issues, even if programmers替代后可转岗, but过于激进society and相关人员未必ready, U.S.推进可能more激进, China will温和很多. U.S.成本很高, electricity, computing power建立在高成本上business model, currently本质上not赚钱, Anthropic and OpenAI年度都不赚钱,成本太高. Silicon-based inflation明显, especially upstream inflation, and silicon-based替代carbon-based明显. But silicon-based inflation与carbon-based通缩contradiction—silicon-based最终要carbon-based买单, cycle到一定程度会出问题. Apple price increase就是market worrying about this problem. China's AI model更能harmonious, commercialization跑通, silicon-based通缩, carbon-based复苏才是正道. Technology应控inflation,以人为本, help碳基做大蛋糕. China-U.S. models有差异also有correlation,还有competition关系. U.S.最强模型technically still占优, but China's最领先模型差距不大, Musk also thinks China models也许next year一季度追上U.S.最领先模型,差距within一年. China性价比高,能做到相近产出,对U.S. AI构成challenge. So U.S. AI based on silicon-based inflation, carbon-based通缩,替代carbon-based又找carbon-based买单,再面临China competition, its development能否stable持续needs observation.
Investing in AI cannot只看上游, application space很大.
Host: After discussing China-foreign AI differences, next focus on domestic AI chain. You proposed domestic AI chain要"去伪存真"and"缩圈". In the current AI industry chain, which segments are过度定价? Which still have cycle-crossing certainty? Are there量化screening standards for viewers参考?
Chen Guo: This question more complex,涉及short-medium-long term investment. Short-term可能只看景气度,如supply-demand紧张,价格上涨即可,但需评估valuation是否合理, because价格上涨持续性,持续多久,之后是否明显下跌, if下跌则EPS难给,可能出现profit负增长. Theoretically做短期交易没问题,只要在price高点前退出即可. If做配置,要找supply-demand紧张且supply集中在少数几家公司的环节,越少越好—一家远好于三家or五家, even三家可能形成寡头垄断. Additionally,投入门槛需要投入的资金越高越好,经历过以前亏损,越老登越好,经得起周期起伏的公司不会过于自信, supply不易一下子过剩, these比较重要.还有个factor,从中期看,投AI不能只看上游, application的空间很大, but不一定当前市场最热捧的方向.
Comments