Inflection Point: When capital shifts first, the market structure begins to diverge

In the official communique of the second-quarter regular meeting of the Monetary Policy Committee of the People's Bank of China released on July 8, a new phrasing for analyzing issues and challenges has been added — structural divergence.

The communique states that China's economic operation is generally stable, advancing toward new growth drivers and higher quality, with new achievements in high-quality development, but it still faces issues and challenges such as strong supply amid weak demand, structural divergence, and external shocks.

These four newly added characters are like drawing two distinct slopes beneath a stable aggregate economic curve: one is "advancing toward new growth drivers and higher quality," where technology, exports, and advanced manufacturing continue to rise; the other remains dragged down by the real estate sector, domestic demand, and the outdated balance sheet structure.

Ming Ming, chief economist at CITIC Securities, links "advancing toward new growth drivers and higher quality" to the development of new productive forces and the optimization of the growth structure, while attributing "structural divergence" to the uneven recovery of manufacturing prosperity, noting that traditional sectors like real estate and infrastructure are still bottoming out. Dong Ximiao, chief economist at China Merchants Union Consumer Finance, believes that the second-quarter regular meeting reflects a shift in macroeconomic regulation: from prioritizing short-term growth stabilization to balancing short-term growth stabilization with long-term structural optimization.

In other words, the People's Bank of China has for the first time formally included the disparities hidden beneath aggregate average figures into its official list of economic issues.

This phenomenon is not unique to China. Almost simultaneously, the minutes of the first interest rate meeting chaired by newly appointed Federal Reserve Chair Walsh also formally incorporated AI (artificial intelligence) into the unified framework analyzing growth, inflation, and interest rates. While the People's Bank of China describes the situation as "structural divergence," the Federal Reserve is discussing AI-related capital expenditure, productivity, and price pressures. Behind these two sets of terminology lies the same core shift: the total economy is still growing, but the sources of growth, associated costs, and the groups benefiting from it are no longer uniform.

Then, which industries have already shifted to a new growth track, which sectors are still trapped in the old economic cycle, and where is capital flowing?

Capital Turns First

Where does capital come from, whose balance sheet does it enter, and what price changes does it drive?

Following the flow of funds, three channels — government credit, bank credit, and equity capital — are opening simultaneously, but they are not flowing evenly to all sectors.

First is government credit. In the first five months of 2026, the cumulative increment in China's total social financing reached 17.48 trillion yuan, of which net government bond financing amounted to 5.67 trillion yuan, accounting for roughly one-third of the total. When financing demand from the private sector is weak, the public sector takes the lead in expanding its balance sheet, converting savings in the financial system into tangible projects, equipment, and orders.

The divergence felt by banks is even more intuitive. Over the first five months, loans to enterprises and public institutions increased by 9.63 trillion yuan, while household loans decreased by 631.4 billion yuan, and household deposits rose by 5.63 trillion yuan.

Earlier released data on loan allocation for the first quarter shows that loans to technology-based small and medium-sized enterprises maintained double-digit growth, while real estate loans and individual housing loans were contracting.

Combining these sets of figures paints a clear balance sheet picture: the government and technology enterprises are expanding their balance sheets, households are leaning toward higher savings, and real estate-related credit continues to shrink.

The third channel is equity capital. The National Venture Capital Guidance Fund has moved beyond its initial establishment goals to operate regional funds, sub-funds, and direct investment projects; relevant policies are also exploring the use of merger and acquisition funds to improve the exit mechanism for investments. Unlike bank loans, this model does not increase enterprises' debt burden. Instead, public funds take on the risks of earlier-stage, longer-term projects first, then attract social capital to enter the fields of R&D, equipment manufacturing, and industrialization.

Resources are concentrating in emerging industries, and policies also need to guard against this concentration evolving into redundant construction, price wars, and low-efficiency production capacity.

To a certain extent, the current inflection point in capital flows is more accurately described as government credit taking the lead, followed by bank credit, with equity capital being guided to flow in, while the capital market has already started pricing these shifts in advance.

Moreover, the demand for AI-related capital expenditure is so large that it cannot be supported solely by venture capital investments and the internal cash flow of technology companies, leading to a rapid surge in debt financing for this sector.

Morgan Stanley estimates that global AI-related bond issuance in 2026 could reach approximately $570 billion; as of May 31, roughly $236 billion of such bonds have been issued, four times the amount in the same period last year. The total annual AI-related expenditure of Alphabet (Google's parent company), Amazon, Microsoft, and Meta is expected to reach around $700 billion.

The concentration of capital in emerging industries does not mean that the lower branch of the K-shaped economic trajectory will recover naturally. Xing Ziqiang, chief China economist at Morgan Stanley, focuses on the challenges facing this lower branch. He points out that while advanced manufacturing and green industries form the rising end of the trajectory, the real estate sector and the old economy are larger in scale and support more jobs and consumer activity. Therefore, while continuing to advance the development of hard-core technologies, China also needs to implement more forceful measures to restore domestic demand.

How Does the Market Price These Shifts

The structural diversion of capital in the real economy and the uneven momentum of industrial capital expenditure will eventually be reflected in the capital market, potentially driving valuation divergence across different stock market sectors, variations in capital crowding levels, and even correlated price fluctuations across different markets.

On July 13, South Korea's KOSPI index dropped by more than 8% during intraday trading, triggering a market circuit breaker; SK Hynix's share price fell by 15.4%, while Samsung Electronics' share price declined by roughly 10%; Japan's Nikkei 225 index also dropped by 2.2%. At the same time, escalating conflicts in the Middle East pushed up oil prices and bond yields, creating a day where profit-taking in the semiconductor sector, cooling earnings expectations, and leveraged trading activities converged to amplify market volatility.

According to the latest report from BofA Global Research, its Bull & Bear Indicator has risen to 9.5, entering a contrarian "sell" zone; in its previous global fund manager survey, 80% of respondents viewed "long positions in semiconductors" as the most crowded trade in the market.

Michael Hartnett, chief investment strategist at BofA Global Research, describes Japanese bank stocks as the "canary in the coal mine" for global markets: if yields on Japanese long-term government bonds continue to rise and begin to pressure bank stocks that previously benefited from interest rate normalization, the low-yen-cost financing model and global carry trade could face disruptions.

This adds the final layer of meaning to the concept of the "inflection point." While where capital comes from and where it flows is undoubtedly important, the cost of capital is equally critical. AI-related capital expenditure can continue, and technology sector credit can keep expanding, but both rely on several preconditions: no hard landing of the economy, no further monetary policy tightening by the Federal Reserve, no cuts in investment by cloud service providers, and continued financing support from the capital market. A change in any of these conditions could cause the upward slope of growth to reverse first on the financing side.

An inflection point is often not the moment when all market participants see the same direction, but rather the moment when everyone has already positioned themselves on the same side of the trade.

Asset management institutions have begun to incorporate tiered allocation strategies into their investment frameworks. Monica Defend, head of the Amundi Institute, believes that central bank independence, inflation volatility, and rising market concentration are jointly shaping a new investment environment, requiring portfolios to enhance resilience through cross-currency allocation, investment in real assets, and holdings of gold. Vincent Mortier, group chief investment officer at Amundi, points out that the AI narrative has shifted from a race to develop cutting-edge models to a focus on who can deploy the technology at scale, with opportunities spreading along the value chain to energy, infrastructure, equipment manufacturing, and real-economy applications.

However, capital inflow does not guarantee locked-in returns. Yao Yuan, senior investment strategist for Asia at the Amundi Institute, warns that large cloud service providers are facing mounting pressure on their free cash flow, meaning future capital expenditure will rely more heavily on bond issuance and equity financing.

Li Changfeng, head of market strategy at AllianceBernstein, offers a reminder from the valuation perspective: for highly cyclical technology hardware sectors, low valuations sometimes correspond to the peak range of earnings expectations. Some AI assets that appear not expensive may simply have their profit denominators inflated by overly optimistic earnings forecasts, lowering forward price-to-earnings ratios. Once the upward revision of earnings slows down, their valuations will be recalculated.

Therefore, AllianceBernstein does not recommend exiting the AI sector entirely. Instead, it suggests retaining hardware assets with high earnings realization certainty, while increasing holdings of non-AI assets with stable cash flow and reasonable valuations, using high-dividend sectors like banking to balance the high volatility of a single industry track. Li Changfeng believes that the strong performance of the Asia-Pacific market in the first half of the year was highly dependent on upward earnings revisions in the AI sector. As the growth rate of overseas AI capital expenditure approaches the peak of this cycle, the market has entered a phase of overheated expectations and intensified valuation divergence. That said, China's export upgrading, booming memory industry, and domestic substitution trend still provide relatively independent fundamental support for some local technology assets.

From the perspective of hard assets, BlackRock's Chief Investment Office observes a multi-year expansion wave for data centers, power facilities, and related infrastructure. BlackRock estimates that by the end of 2026, its own investments and committed funds in data centers will reach approximately $100 billion; together with roughly $200 billion in third-party investment in AI chip production capacity, near-term investment in AI infrastructure will total around $300 billion — a scale equivalent to the GDP of an economy ranking among the top 50 globally. BlackRock regards power supply, land resources, long-term customer contracts, and strong execution capabilities as the key factors that distinguish high-quality projects from unconstrained, low-efficiency expansion.

These three lines of observation all focus on one core question: capital is indeed concentrating in hard-core technology sectors, but this concentration does not guarantee that all related assets will keep rising in value. Capital expenditure is still expanding, but financing conditions are now determining its growth slope; AI infrastructure remains scarce, but project quality is diverging into different tiers; AI may boost productivity, but the profit and cash flow of different links in the industrial chain have already moved in different directions.

Translating this to the Chinese market, Yao Yuan does not propose a unilateral bullish view, but a "structurally neutral" strategy: overweighting industries representing new productive forces, while underweighting old-economy sectors such as consumer services and real estate.

The so-called inflection point may not be an overall downturn in the AI cycle, but a redistribution of capital, production capacity, and profits within the cycle itself.

In 2010, Gao Shanwen, former chief economist at SDIC Securities, studied the Lewis Turning Point by identifying signs of labor shifting from surplus to shortage through micro-level data series such as agricultural product prices, low-skilled labor wages, and service prices. His core concern was where macroeconomic changes would first leave their traces in price movements.

This divergence is not limited to the gap between technology and traditional industries; it also exists between the profit performance, financing conditions, and share price movements within the same booming industry cycle.

Following this analytical approach, today we cannot only focus on AI-related slogans and index fluctuations, but also examine whether capital expenditure, financing costs, free cash flow, and return on capital are all simultaneously shifting their growth slopes.

Lu Lei, vice governor of the People's Bank of China, reaffirmed Gao Shanwen's judgment in an article commemorating the economist: for monetary authorities and financial markets, what is truly valuable is not the trend (which everyone can easily see and understand), but the inflection point — which is elusive but determines the success or failure of economic decision-making and investment.

Today, this inflection point has taken on a new form: the total economy remains stable, but different industries are heading toward completely distinct growth trajectories.

Perhaps the inflection point in this non-linear economic era is no longer a moment when all market participants turn around in unison, but a point where structural divergence begins to accelerate.

This article is sourced from the WeChat Official Account "Economic Observer", authored by Ouyang Xiaohong, and published by 36Kr with official authorization.