As the initial excitement fades, the market rebound is likely transitioning into a plateau phase. The sharp decline in global AI assets in July was driven by multiple factors, including heightened concerns over fundamental contradictions within the industrial chain, geopolitical risks from the US-Iran conflict, and rising expectations of liquidity tightening. Since August, renewed signals of US-Iran negotiations, along with weaker US employment and inflation data post-World Cup that eased expectations of tightening by the Federal Reserve, have led to a gradual recovery in tech assets. However, the risks of industrial chain contradictions and excessively high fundamental expectations persist. The phenomenon of "earnings exceeding expectations but stock prices falling sharply" has occurred repeatedly. Recent rebounds in tech assets, driven by fundamental improvements, have largely been structural or have temporarily eased concerns about specific links in the industrial chain, rather than reversing market expectations or ushering in a new round of trend-driven sector-wide momentum. As the upward momentum from liquidity weakens, the comprehensive rebound of tech assets may be entering a bottleneck. High interest rates have been more favorable for tech stocks, but declining rate levels could also drive a rebalancing of market styles in the US stock market. In the A-share market, a similar rebalancing trend is evident, with non-AI sectors such as healthcare, real estate, and food and beverages delivering notable gains. Within the tech space, divergence is emerging, with communication stocks outperforming while electronics show signs of stagnation. It is worth noting that the current AI industrial cycle is dominated by the United States. The driving factors behind the rebound in the US AI supply chain, including the commercialization of AI applications, the initial closure of the cloud capital expenditure expansion and commercialization path, the capacity expansion of US optical module manufacturers, and US government restrictions on imports from Chinese optical module makers, all indicate efforts by the US government and enterprises to redistribute profits and repatriate funds within the AI supply chain. This includes large-scale supply agreements between Samsung, SK Hynix, and US tech giants, as well as increased investment by TSMC in the US, all of which deepen ties with the US AI supply chain. In contrast, due to geopolitical constraints, many links in the AI chain between China and the US have already been significantly decoupled, while closely connected overseas computing power chains are suppressed by US import policies. This has prevented the narrative of the US tech stock rebound from being directly mapped or transmitted to the A-share market. In fact, as the "K-shaped divergence" in export sectors intensifies, AI-related exports have led to a decline in exporters' willingness to settle foreign exchange, and overseas asset scales have begun to expand again.
The mismatch between debt and income suggests time may not be on the side of AI assets. As AI capital expenditure shifts from internal cash flow to external debt financing, market attention is moving from "expansion of AI investment" to "validation of AI investment returns." On the demand side, after the rapid adoption of AI agents like Claude Code and OpenClaw in the second half of 2025, corporate willingness to pay for AI products has only recently begun to accelerate significantly. As of now, the core demand for AI products from enterprises remains centered on API calls (accounting for over 50% of AI product spending). This implies that the narrative supporting debt expansion in the AI supply chain primarily relies on the nonlinear growth of the Token market size. The ideal commercialization path is: downstream agent penetration and diffusion — growth of Token revenue (market size) — continued supply-demand imbalance for computing power — further increases in capital expenditure by cloud and large model companies — and the rationality of large-scale debt expansion. Therefore, if the expansion of Token market size lags behind debt growth, leading to signs of rising leverage ratios in the supply chain, market concerns about debt pressure will intensify. Since 2026, leverage ratios have shown a downward trend overall, and the clarity of the commercialization path has provided a tailwind for tech stock gains. However, since mid-to-late July, we have observed a re-increase in leverage ratios. If this trend continues, market concerns about debt pressure will intensify, constraining the ability of cloud companies to further expand debt. Large model and cloud companies themselves will also reassess the alignment between capital expenditure and returns. On the supply side of funding, non-bank financial institutions (e.g., VC/PE), as important marginal sources of capital for AI and cloud computing capital expenditure, have seen commercial banks' credit support for them stop increasing. This may signal that conservative investors are becoming wary of debt pressure in the AI supply chain.
The bottom for real assets is emerging. For real assets or HALO assets, AI investment remains at relatively high levels, the decline in real interest rates is limited, and the US economy is only slowly slowing from high levels—far from needing interest rate cuts. The Strait of Hormuz is still closed, shipping volumes are recovering slowly, and a full recovery in demand from non-US economies, especially emerging markets, may still require time. However, while AI needs to prove it will usher in a new round of expansion, time is on the side of real assets, and the bottom has been gradually identified. For commodities, as real interest rates begin to decline and converge with inflation expectations, their equal-weighted prices have started to bottom out and recover. Looking ahead, as interest rates and the US dollar weaken, the gap between the US and non-US economies, as well as between tech assets and HALO assets, will narrow. Referring to the period from Q4 2025 to Q1 2026, when the AI leverage ratio rapidly rose to high levels and US Treasury real yields declined, HALO assets began to show higher flexibility compared to the broader market returns.
During this period of market tug-of-war, attention should be focused on changes beyond the "attention assets." As the repair driven by the denominator side (e.g., liquidity) gradually concludes, and the numerator side (e.g., fundamentals) has yet to see systematic major catalysts or is in a tug-of-war, both AI assets, which are currently the focus of market attention, and non-AI-related real assets have entered a plateau phase. However, looking ahead, the advantage of AI-related assets is their current high-level prosperity, while unfavorable factors are accumulating. Based on past experience, market investors are more willing to pay an option premium in stock prices, meaning nonlinear changes always appear again. However, this option premium itself will erode over time. For the broader real asset demand side, which is on the friendly side of time erosion, the bottom for fundamentals and asset prices is being confirmed. If the ratio of new debt to income in the AI supply chain continues to rise, it will significantly constrain the financing capabilities and capital expansion of cloud and large model companies. As the economic fundamentals of the US and non-US economies converge, similar to the scenario in late 2025 and early 2026 when the AI leverage ratio rose and US Treasury real yields declined, real assets will exhibit flexibility. In terms of specific allocation recommendations: First, the rebound in commodities has been established, with the energy plus metals combination transitioning from a downward trend since March to an upward phase. The US-Iran conflict only determines the internal distribution. Considering the short-term intensification of the conflict, the current resource ranking is: energy (coal, oil, and petrochemicals), non-ferrous metals (gold, copper, aluminum). Second, the dividend style benefits from the rotation of absolute return investors. High-dividend, low-volatility, and stable cash-flow dividend assets remain the core direction for capital return. Third, the resonance of the "Southern countries" story and China's manufacturing is incubating, with attention on areas such as construction machinery, power grid equipment, and refining.
Risk Reminder: If the domestic economic recovery falls short of expectations, the assumptions in the article about capital market improvement from fundamental improvements would not apply. If the overseas economy experiences a significant downturn, the global manufacturing joint recovery may pause, and real asset demand could slow down.
Comments