According to a recent report from CICC, global stock markets have experienced a synchronized pullback since July, led by the AI sector, intensifying debate over the bull market's sustainability. This follows a strong rally in the first half of 2026. The firm argues that the core of this correction is the shift of three key variables—the AI industry trend, USD liquidity, and the Chinese economy—from a previously supportive "Triple Support" to a current challenging "Triple Constraint," prompting a reassessment of earlier optimistic assumptions.
The "Triple Support" in the first half fueled market gains. At the start of the year, consensus favored a continued bull market for global and Chinese stocks, but significant disagreement revolved around three variables: whether the AI trend was a bubble or a revolution, the policy stance of the incoming Fed Chair, and the trajectory of the Chinese economy. These variables formed a favorable combination in H1. For AI, breakthroughs in enterprise-facing AI agents led to exponential growth in annualized recurring revenue for companies, suggesting viable profit models. On USD liquidity, the US-Iran conflict initially spiked oil prices, but subsequent truces and a memorandum of understanding led to a sharp decline in oil and improved inflation expectations. For China, the economy saw a strong start with robust export growth, the GDP deflator ending 13 consecutive quarters of decline, and Q1 A-share corporate earnings reaching their best in five years. Supported by these three factors, A-shares and global markets rallied, with indices having higher AI exposure showing stronger performance.
The shift to "Triple Constraints" has been stark. Since July, the AI sector has led a global market pullback, with Chinese indices erasing most of their H1 gains, moving from global leaders to laggards. This is because the supportive variables have reversed. Doubts about AI investment returns and bubble fears have resurfaced, overseas liquidity tightening expectations have intensified, and traditional sectors of the Chinese economy face downward pressure. This creates challenges for both the numerator (earnings) and denominator (valuation) of domestic stocks, spanning both traditional and new economy sectors. However, historical experience suggests markets often overestimate short-term shocks while underestimating long-term structural trends, as investors tend to overweight extreme near-term events. Therefore, it is crucial to analyze the logic behind these "Triple Constraints" to determine if the shocks represent a trend reversal or just a temporary pause.
AI Industry Trend: From Narrative to Return Validation
The essence of the wobble in the AI narrative is a fundamental questioning of AI investment returns and the sustainability of capital expenditure. The global adjustment in the AI sector began in late June, with negative narratives echoing past concerns about investment returns. During the July earnings season, while the capital expenditure guidance for the top five North American cloud providers was still being revised upwards, with 2027 capex forecasts approaching $1 trillion, concerns are shifting. The ratio of capex to operating cash flow for these companies reached 97% in Q2, and Google's free cash flow turned negative. The market's focus is moving from input scale to return validation. Any negative news regarding revenue generation from AI models can quickly reignite concerns about capex sustainability. Several specific negative developments since late June have acted as catalysts for the decline. Apple's price increases for iPads and Macs, due to rising memory costs, signaled that upstream cost increases are passing through to end-users. More importantly, the market realized that broad upstream price hikes erode the output per unit of capital expenditure, potentially lengthening payback periods for cloud providers if token output efficiency doesn't improve simultaneously. The release of the Kimi-K3 model was a significant event, achieving top global rankings in benchmarks and sparking another "DeepSeek moment." The model's competitive pricing, significantly lower than comparable overseas models, intensifies competition between Chinese and US AI models, pressures unit pricing for equivalent performance, and exacerbates concerns about the return on investment for overseas models. Furthermore, market concerns about circular financing within the AI industry chain escalated after news broke, leading to panic and a noticeable widening of CDS spreads for the top five cloud providers. The amplitude of the correction was amplified by leverage and high market crowding. The global rally in the AI sector in H1 was significantly aided by leverage, notably in the Korean market where credit financing balances surged. The massive issuance of single-stock leveraged ETFs, particularly for Samsung and SK Hynix, created a vulnerability where ETF rebalancing during a downturn can trigger a further downward spiral. In China, while A-share margin balances also hit a record high, the concentration of funds in AI hardware significantly increased crowding risk. Domestic actively managed equity funds saw their allocation to electronics and communications reach 60% in Q2, a degree of concentration surpassing four previous "herding" episodes. This combination of leverage, high crowding, and large IPOs absorbing liquidity amplified the impact of negative AI narratives, leading to significant valuation compression across global AI hardware markets, even as earnings expectations remained largely unchanged. The Korean stock market and the Philadelphia Semiconductor Index experienced their largest monthly declines since the financial crisis, while the ChiNext index saw its second-largest monthly decline on record.
Global Liquidity: From Rate Cut Expectations to Rate Hike Expectations
The volatile US-Iran situation has reshaped the inflation path and rate hike expectations. The initial conflict and subsequent truce caused oil prices and US CPI to fluctuate. However, the Fed did not turn dovish with the interim oil price decline. When tensions flared again in July, restricting passage through the Strait of Hormuz and pushing oil prices higher, rate hike expectations quickly intensified. From an initial market expectation of three rate cuts in 2026, the market now prices in at least one rate hike for the year, even after weaker-than-expected non-farm payroll data. This shift from "rate cut trading" to "rate hike trading" forces a repricing of valuations that were previously inflated by liquidity. Rising global bond yields and tightening liquidity have become a source of market pressure. Ten-year government bond yields in the US, Germany, Japan, and the UK have risen concurrently. The US 10-year and 30-year yields have climbed, reflecting deep-seated concerns about long-term inflation and fiscal debt. For the AI industry, which requires significant borrowing, higher financing costs raise the required rate of return on investment, further impacting valuations.
Chinese Economy: The "K-shaped" Divergence Reduces Tolerance for Risk
The Chinese economy displays a distinct "K-shaped" divergence, with emerging industries outperforming traditional ones and external demand far exceeding internal demand. The AI sector has been a key growth driver, but traditional sectors are under pressure from a persistent property downturn and relatively slow fiscal spending in the first half. This K-shaped divergence has been priced into the stock market for over a year. Since September 2024, the AI-related sector index has surged roughly 200%, while the non-AI sector index has only risen about 34%. This significant outperformance of the booming sector, combined with the weakness in the traditional economy, means that the tolerance for error in the booming sector is low. Any negative catalyst can lead to outsized declines in the broader index.
While the "Triple Constraints" have altered the market's rhythm, whether the trend has ended depends on the future evolution of industry trends, USD liquidity, and the Chinese economy. CICC believes the current sharp correction is heavily linked to the extreme market divergence and excessive trading crowding. The bearish sentiment during the correction has been overly amplified. The firm maintains a positive outlook on the AI industry trend as a base case, and argues that the market may be overestimating the risk of Fed monetary tightening. The Chinese economy's fundamentals also have the conditions to improve. Therefore, the "Triple Constraints" may not necessarily lead to the pessimistic scenario the market fears. Supported by the "New Order, New Drivers, and New Ecosystem" that the A-share market has established, a "slow bull" market remains the base case scenario.
AI Industry Trend Unchanged: Not Yet at Bubble Bursting Stage
Assessing the risk of an AI bubble bursting involves three dimensions: productivity improvement, debt leverage risk, and pricing irrationality. The productivity gains from AI are undeniable, and the current market focus on return on investment does not yet meet the criteria for a levered, irrational bubble. The market may be underestimating the growth in AI demand and the performance improvements of models and hardware. Token usage on major platforms has grown nearly 20-fold year-on-year, a trend likely to continue with the spread of AI agents and applications. From a pricing perspective, as AI evolves from simple Q&A to complex tool calling and agent collaboration, customers are more concerned with faster, more accurate task completion. Consequently, token pricing can actually increase with improved model performance, as seen with recent model releases. While long-term price competition is inevitable, the ability to set higher prices for superior performance, coupled with hardware efficiency gains, can offset rising costs. The leverage of leading companies remains low, and overall valuations are not yet at the levels seen during the dot-com bubble. The debt-to-equity ratio for the top five cloud providers is around 43%, significantly lower than the ~124% average during the internet revolution. Their financial leverage is healthy, with investment primarily funded by internal cash flow, not high-leverage expansion. The key risk is not debt itself, but low-return projects funded by high leverage, a scenario not currently prevalent. The forward P/E ratio for the S&P 500 Information Technology sector is about 21x, down from end-2023 due to fast earnings growth, and well below the ~55x peak of the dot-com bubble. Overall valuations for leading AI companies are relatively healthy, though some second-tier companies with lower barriers to entry may have experienced some valuation froth. Historically, major tech bull markets experience multiple deep corrections before their final peak. The core determinant is whether the underlying industry trend persists. As seen in the dot-com bubble, the Nasdaq experienced four significant pullbacks before its 2000 peak, the largest of which was over 25% during the LTCM crisis. The index eventually recovered and surged 150% higher because the industry trend remained intact. Therefore, the current pullback should not be automatically interpreted as the end of the AI bull market. As long as technological progress, demand, and corporate earnings continue to reflect a positive trend, the AI bull market is likely to continue, though internal differentiation and rotation are expected.
Current Environment Does Not Support Fed Monetary Tightening
US inflation is expected to enter a downward trajectory in the second half of the year, and economic growth is cooling, which does not support monetary tightening. While the US-Iran situation has worsened, the probability of a policy de-escalation remains, given the upcoming mid-term elections. The one-off impact of energy prices on inflation is fading, and leading indicators for rents, used cars, and other core goods point downwards. The central bank's decision to raise rates depends on whether the oil price shock triggers "second-round effects," which CICC believes is unlikely. Cooling labor market data and the substitution effect of AI are expected to further suppress wage and services inflation, preventing a wage-price spiral. Furthermore, the widening growth divergence between the AI and non-AI sectors of the US economy does not support tightening. The current Fed Chair, perceived as a hawk, may be more dovish in practice. The Fed did not raise rates at the July FOMC meeting, and the Chair's comments suggested that the significant rise in long-term bond yields has effectively done the Fed's tightening for them. The creation of five working groups, including one on data sources focusing on trimmed mean inflation (which is less sensitive to volatile supply shocks), and another on productivity and employment that may incorporate AI into the policy reaction function, suggest a more nuanced approach. The historical correlation between major technological revolutions, productivity gains, and falling inflation provides a reference for a scenario of higher growth with lower inflation. CICC believes the market may be overestimating the risk of monetary tightening, and with inflation set to fall and growth weakening, the Fed's policy has room to return to an accommodative stance in the second half of the year.
Domestic Policy Has Ample Room, Economy Likely to Remain Resilient
China's policy space remains ample, and H2 efforts are likely to stabilize the economy. The late July Politburo meeting sent positive policy signals. It acknowledged the challenges facing the economy while highlighting the positive trend of "new momentum and structural optimization." Crucially, it reiterated the need to "increase counter-cyclical adjustment" and "promptly plan practical and effective incremental policies." On fiscal policy, the meeting specifically called for "accelerating the pace of fiscal expenditure and bond fund usage," indicating that the government has the space but has chosen to delay its efforts. CICC believes that accelerating fiscal spending and project implementation will help stabilize economic fundamentals in H2. The microeconomic foundation of the current economy is also better than in the past three years. The earlier period of low inflation was partly due to supply-demand imbalances from rapid capacity expansion. However, as various sectors begin to cut capacity, the firm sees a clear decline in A-share manufacturing capital expenditure, with the ratio of capex to depreciation dropping to a historical low. This is gradually leading to a decline in projects under construction, and the indicator of capacity utilization (revenue to fixed assets + construction in progress) is stabilizing. The sequential improvement in PPI before the recent oil price shock reflected the positive impact of this capacity reduction on prices. Therefore, with an appropriate policy response, the medium-term economic fundamentals are expected to show considerable resilience.
Slow Bull Market Remains the Base Case Scenario
Based on the analysis above, the current "Triple Constraints" do not represent a trend reversal. CICC believes they could turn into supportive factors over time, allowing the market to regain its upward momentum. The firm's "Three News" logic supporting a slow bull market for A-shares remains fundamentally unchanged. The "New Order" refers to the accelerating structural transformation of the international monetary system, where the direction of capital flows is more powerful than short-term national fundamentals. This is a key reason to be bullish on Chinese assets. The core of this transformation is the declining safety of USD assets, driven by unsustainable US government debt and a shift towards more transactional, national-interest-first policies that undermine the neutrality and predictability of the dollar system. Despite recent market discussions about the "end of de-dollarization," CICC argues that the trend is far from over. The persistent rise in US government debt continues to erode the "safe asset" consensus around Treasuries, and the structural differences between equities and bonds mean US stocks cannot fully replace the reserve asset function of Treasuries. The AI revolution, combined with a potential "reverse Balassa-Samuelson effect," does not necessarily strengthen the dollar's position. This international order shift will drive global capital flows towards "fragmentation and diversification," and Chinese assets stand to benefit from a long-term revaluation of their allocation value. The "New Drivers" are the transformation of the economy, with the tech industry emerging as a new growth engine. The A-share market's sector composition is already reflecting this shift, with AI, new energy, and high-end manufacturing now accounting for roughly 60% of the index weight, while traditional sectors like banking, oil, and coal account for only 15%. This provides the A-share market with better growth potential, creating conditions for a medium-to-long-term slow bull market. The "New Ecosystem" refers to a more balanced supply and demand dynamic in the market, thanks to positive institutional changes. The "New Nine National Guidelines" have encouraged companies to increase dividends, and listed companies' free cash flow is at a historical high, leading to a significant increase in payouts. The total buyback and dividend distribution on A-shares has already exceeded the amount raised from IPOs and refinancing. Furthermore, the first year of accelerated entry for long-term funds has seen insurance companies increase their equity holdings above their historical average. The "National Team" (Central Huijin) has a mature market stabilization mechanism, and new regulations encourage mutual funds to focus on long-term performance. These multiple forces are creating a positive cycle of capital inflow, which is more sustainable than in the past.
Market Characteristics and Coping Strategies During the Verification Period
The verification period is characterized by a "focus on reality over narrative" and increased market volatility. While CICC believes the "Triple Constraints" may ultimately prove to be a false alarm, the three supporting assumptions need to be re-proven. During this period, market decisions are driven less by narrative extrapolation and more by the tangible realization of earnings and cash flow. This period also brings lower tolerance for error, leading to higher implied volatility, as seen in the ChiNext and CSI 1000 indices. The difficulty of investment is expected to be higher than in H1. The recommended strategy for this period is to shift from broad-driven market gains to prudent stock selection. For Chinese stocks, the base case for a slow bull market remains intact, but the upward slope is expected to be more gradual as the market waits for validation. At the sector level, valuation expansion potential should be treated with greater caution, with a greater emphasis on bottom-up industry logic and a more balanced style. Within the AI sector, further differentiation is expected, with performance broadening from the "picks-and-shovels" hardware plays to model and application layers. For infrastructure companies, those with strong technological barriers and high order visibility are expected to regain their upward momentum, while those with lower barriers may need to digest high valuations. Key areas of focus include optical communication, domestic AI chips, leading large model companies, and downstream AI applications. Beyond AI, the firm recommends looking at energy transition, cyclical sectors with improving capacity utilization and export demand, and assets sensitive to improving USD liquidity. Additionally, high-dividend stocks with stable free cash flow can serve as a buffer against volatility in growth assets. In terms of asset allocation, CICC is bullish on US and Chinese equities, and overweight on gold and non-ferrous metals. The recent global correction has made valuations more reasonable, and the AI sector's fundamentals are still being delivered. With the risk of leverage and crowding partially digested, the firm expects both US and Chinese markets to perform well in H2. On gold, the firm recommends an overweight position, as the bearish narratives are being disproven. The global liquidity environment is not set for a sustained tightening cycle, and the "de-dollarization" trend is not ending, as evidenced by strong central bank gold purchases. Gold is expected to benefit from both improved liquidity and the ongoing diversification of monetary reserves. For US Treasuries, a neutral position is recommended, as falling inflation and a cooling labor market are positive for rates, and high coupons provide a safety buffer. Within commodities, the outlook is for continued divergence. Metals like copper and aluminum, supported by AI data centers, grid expansion, and energy transition, are recommended for overweighting. Energy commodities like oil maintain hedging value but are expected to see increased volatility. Finally, the firm recommends an underweight position on Chinese government bonds, as the potential for further yield declines is limited given the already low interest rate levels. Key risks include AI capital expenditure returns falling short of expectations, a surprise tightening in global inflation and monetary policy, further escalation of the US-Iran situation pushing up energy prices, and weaker-than-expected domestic economic and policy outcomes.
Comments