From a growth perspective, the market is focused on the artificial intelligence, or silicon-based economy, theme. Furthermore, capital is highly concentrated in upstream infrastructure hardware chains that benefit from AI capital expenditure, essentially the upstream silicon sector. We believe a key issue to monitor going forward is the balance between the upstream silicon sector, downstream silicon applications, and the carbon-based economy.
We posit that the lesson from the 2000 dot-com bubble burst was fundamentally about excessive capital expenditure in the upstream internet economy, while the downstream internet economy, as a whole, had not yet established a sustainable and stable profit model through effectively empowering the real economy. This phenomenon is likely to recur with each wave of productivity revolution, known in economics as the productivity J-curve effect. Economists Erik Brynjolfsson, Daniel Rock, and Chad Syverson argue that in the early stages of technological progress, significant resources are often invested in unmeasured intangible assets, while output has not yet materialized, leading to statistics showing declining or stagnant productivity. The later rise occurs when these intangible assets begin to generate returns, and output growth appears. However, capital is often overly optimistic during the initial phase of technological advancement when overall economic productivity is declining or stagnant, leading to a tendency for premature overinvestment. In the U.S., some indicators warrant caution. For instance, the U.S. ICT investment-to-GDP ratio in Q1 2026 has already surpassed the peak levels seen during the 2000 dot-com bubble. The subsequent focus should be on the risk that U.S. ICT investment and stock market valuations for tech companies may outpace the development of the real, or carbon-based, economy.
Recent signals indicate progress in U.S.-Iran negotiations, increasing the probability of an agreement. This could lead to a marginal decline in U.S. Treasury yields, which have been suppressing market valuations, potentially providing further upward momentum for the AI tech chain. However, on the other hand, as strategic petroleum reserves and inventories have been significantly depleted over the past months, some Middle Eastern oil fields have sustained damage from earlier wartime destruction and prolonged shutdowns, and a full restoration of Strait transit rates to pre-war levels is expected to take considerable time, we anticipate that energy price costs, such as for oil and gas, will remain at a relatively high median this year. Furthermore, from an energy security perspective, a new global wave of energy transition is likely to continue.
Key sectors to watch include: the AI industrial chain (PCB, MLCC, semiconductor equipment, CPUs, etc.), energy security (wind, solar, energy storage, vehicles, grid equipment, power equipment, coal, oil & gas, etc.), humanoid robots, commercial aerospace, machinery, and innovative drugs.
Risk factors include: intensifying overseas risk disturbances, weaker-than-expected sustainability of capital expenditure in the AI industrial chain, and domestic policy and economic recovery falling short of expectations.
1. Negotiations Show Progress, but Full Recovery Takes Time Recent signals indicate progress in U.S.-Iran negotiations, increasing the probability of an agreement. However, the median oil price is likely to remain elevated because strategic petroleum reserves and inventories have been significantly depleted over the past months, some Middle Eastern oil fields have sustained damage from earlier wartime destruction and prolonged shutdowns, and a full restoration of Strait transit rates to pre-war levels is expected to take considerable time. We anticipate that energy price costs, such as for oil and gas, will remain at a relatively high median this year. Moreover, from an energy security perspective, a new global wave of energy transition is likely to continue. The recently released Federal Reserve meeting minutes explicitly identified Middle East geopolitical risks as a core driver of persistently high inflation, marking a potential shift in the Fed's policy path from expectations of rate cuts to the possibility of hikes. The market is currently engaged in a fierce tug-of-war between peace premiums and inflation stickiness. From a geopolitical standpoint, U.S. Secretary of State Rubio stated in a May 22 interview in Miami that the U.S. prefers reaching an agreement but that other options are available for the President. While U.S. Treasury yields, which have been suppressing the market, may see a marginal decline, the Fed's FOMC meeting minutes released on May 20 conveyed a clear hawkish signal. The minutes showed that staff projections for 2026 inflation were significantly higher than in the March meeting, primarily due to rising energy prices and supply chain pressures stemming from the Middle East conflict. Most officials acknowledged that returning inflation to the 2% target would take longer than previously anticipated. Participants generally agreed that Middle East geopolitical tensions could significantly impact the economic outlook and policy path, with elevated energy prices continuing to exert upward pressure on overall inflation in the near term. The market is currently caught in a transmission chain where high oil prices support high inflation expectations, which in turn forces a hawkish monetary policy pivot. We believe the dual pressures of oil prices and U.S. Treasury yields are becoming core variables for global asset pricing in 2026.
2. AI Chain's Continued Valuation Re-rating May Increase Overbought Risks; U.S. Indicators Under J-Curve Effect Warrant Caution Rising expectations for a peace deal may lead to further valuation re-rating for tech stocks, but this could increase overbought risks. As of May 22, 2026, the STAR 50, ChiNext, and CSI TMT indices are trading above their respective 20-day Exponential Moving Averages (EMAs), maintaining a short-term bullish pattern. However, the degree of positive deviation is generally at historically high levels, highlighting technical overbought pressure and suggesting potential structural pressure in the near term. From a macro perspective, the three indices have experienced a rapid rally since April 2026, with cumulative gains generally exceeding 20%. While the 20-day EMA, as a medium-term cost line, maintains a steep upward slope, it has failed to keep pace with the price gains, leading to a continuously expanding deviation rate. Historically, this has often presaged consolidation or a phase of correction. Specifically, the deviation for the STAR 50 index is the most extreme. According to data, the STAR 50 closed at 1790.77 points on May 22, 2026, having touched an intraday high of 1832.02 on May 20. The 20-day deviation rate once reached nearly 14%. After a 3.70% single-day drop on May 21, the deviation rate converged to the 7%–8% range, but as of May 22, the closing price remained far above its 20-day EMA from the same period last year. Such a significant distance from the moving average reflects strong momentum but also implies that if market sentiment cools or incremental capital fails to sustain the rally, the index's reversion to the mean could be relatively sharp. The ChiNext index's movement is relatively milder but also in overbought territory. Data shows the ChiNext index closed at 3938.50 points on May 22, 2026, having reached a yearly high of 4038.33 on May 13, with an estimated 20-day deviation rate near 9% at that time. Despite recent volatility, the index has consistently traded above its 20-day EMA. From a longer cycle perspective, the ChiNext index rose approximately 27.76% from around 3160 points on April 7, 2026, to May 13, with the deviation rate failing to correct effectively, indicating significant profit-taking pressure. Compared to the STAR 50, the ChiNext index's deviation is slightly smaller due to its inclusion of more weight in sectors like new energy and pharmaceuticals, but short-term volatility risks persist. The CSI TMT index provides a quantitative reference. Detailed data shows the CSI TMT closed at 3823.0871 points on May 22, 2026, with its 20-day EMA at 3589.33 points, resulting in a positive deviation rate of approximately 6.51%. The TMT index's 20-day EMA continues to rise steeply without a significant slowdown in slope, indicating the medium-term trend remains intact. However, the distance between the closing price and the moving average has failed to expand further for several consecutive sessions, hinting at waning upward momentum. Common characteristics across the three indices include: a medium-term bullish bias, but deviation rates are generally at high historical percentiles over the past year. For a healthy market advance going forward, either sideways consolidation is needed to allow the moving averages to catch up, or an active correction is required to repair the deviation.
The productivity J-curve, proposed by economists Erik Brynjolfsson, Daniel Rock, and Chad Syverson, can explain the initial productivity slowdown and subsequent productivity surge often accompanying general-purpose technologies. The J-curve phenomenon: Initial phase (trough): Because significant resources are invested in unmeasured intangible assets, and output has not yet materialized, statistics show declining or stagnant productivity. Later phase (rise): When intangible assets begin to generate returns, output grows, and previously underestimated costs disappear, leading statistics to show a significant productivity increase. The U.S. ICT investment-to-GDP ratio in Q1 2026 has already surpassed the peak levels seen during the 2000 dot-com bubble. The current massive investment may be in the "initial investment" phase of the J-curve—economic data shows ICT investment hitting new highs, but the return in Total Factor Productivity (TFP) has not fully materialized. According to this theory, these investments (especially in intangible assets) may take several years to translate into broad-based productivity growth.
3. Domestic Economic Divergence: Widening Scissors Gap Between New and Old Growth Drivers Production is stronger than investment, high-tech development outpaces traditional sectors, and the external demand chain outperforms the domestic demand chain. On May 18, the National Bureau of Statistics released economic performance data for April and January-April 2026. Overall, the Chinese economy in January-April 2026 exhibited a pattern of deep divergence: production is stronger than investment, high-tech development outpaces traditional development, and the external demand chain outperforms the domestic demand chain. Industrial value-added maintained relatively fast growth, driven by sectors like computers & electronics, transportation equipment, and automobiles. National real estate development investment growth rate fell again to -13.7% year-on-year. This structural feature reflects both the positive progress in cultivating new quality productive forces and exposes the weakness in domestic demand against the backdrop of a real estate downturn and external demand uncertainty.
3.1. High-Tech Manufacturing Defies the Trend, Post-Real Estate Cycle Remains Weak Observing industrial value-added data, we find that high-tech manufacturing and high-end equipment manufacturing have become core growth drivers. In April, manufacturing of computers, communication, and other electronic equipment surged 15.6% year-on-year, with a cumulative growth of 14.0% from January to April, leading the pack. Manufacturing of railway, ship, aerospace, and other transport equipment grew 8.2% in April, with cumulative growth of 12.1%. Automobile manufacturing grew 9.2% in April, significantly accelerating from the cumulative growth rate (6.1%). These sectors benefit from the global semiconductor cycle recovery, domestic trade-in policies, and export resilience, serving as anchors for current industrial production. In contrast, sectors closely linked to the real estate chain showed continued weakness. Non-metallic mineral products fell 6.5% year-on-year in April, with a cumulative decline of 2.3%. Manufacturing of wine, beverages, and refined tea fell 1.4% in April. Smelting and pressing of non-ferrous metals fell 1.0% in April, indicating soft demand in traditional consumption and the post-real estate cycle.
3.2. Divergence Behind the 1.2% Figure: Strategic Investment Hot, Livelihood Investment Cold Total manufacturing fixed-asset investment grew 1.2% year-on-year cumulatively from January to April, but internal structure diverged significantly. On one hand, strategic emerging industry investment saw substantial growth: investment in railway, ship, aerospace, and other transport equipment grew 24.7%, textile industry investment grew 12.1% (benefiting from export order shifts), and investment in computers, communication, and other electronic equipment grew 5.4%. On the other hand, investment in multiple sectors declined sharply: pharmaceutical manufacturing investment fell 7.8%, special equipment manufacturing fell 6.0%, food manufacturing fell 4.5%, and chemical raw materials fell 3.6%. More notably, investment in education (-11.2%), health and social work (-10.9%), and culture, sports, and entertainment (-9.9%) contracted sharply.
3.3. Macro Dilemma Amid Triple Divergence: Real Estate Drag, External Demand Hedge, and Policy Dilemma Against the macroeconomic backdrop, the aforementioned divergence carries three deep implications. First, the long-term downturn in real estate continues to drag on the economy through two channels: the industrial chain (non-metallic minerals, ferrous metals, etc.) and land finance. The value-added of non-metallic mineral products fell 6.5% year-on-year in April alone. Second, external demand resilience and industrial upgrading are hedging against insufficient domestic demand: the high growth in export-oriented sectors like computers & electronics and transportation equipment partially compensates for weak domestic demand. However, if global demand cools in the second half of the year, this engine may weaken. Third, policymakers face a dilemma: they must continue supporting investment in high-tech manufacturing (e.g., railway, ship, aerospace, and other transport equipment manufacturing, up 24.7% year-on-year from January to April) while also needing to reverse the cliff-like decline in social sector investment. The nearly -11% year-on-year growth in education, health, and social work investment from January to April not only affects short-term employment and livelihoods but also erodes long-term human capital accumulation.
Risk Factors: 1) Global Economic Slowdown Risk: If growth in major global economies, particularly China and the U.S., falls short of expectations, it could lead to a contraction in global demand, negatively impacting commodities, equity, and bond markets, especially in consumer, technology, and cyclical sectors. 2) Commodity Price Volatility: Prices of commodities, especially energy and precious metals, may experience sharp fluctuations due to supply-demand changes, policy adjustments, or international conflicts, affecting related industries and investment returns. 3) Geopolitical Risks: Geopolitical events worldwide, particularly tensions in the Middle East, Europe, and between the U.S. and other major powers, could trigger market volatility, impact risk asset pricing, and further increase global economic uncertainty. 4) Potential errors in data statistics.
Comments