For stock trading, look at the Golden Kirin Analyst Research Report — authoritative, professional, timely, and comprehensive, helping you uncover potential thematic opportunities! Strategy Focus | Coping with High Interest Rates. Source: CITIC Securities Research. Authors: Qiu Xiang, Gao Yusen, Chen Feng, Chen Zeping, Zhang Mingkai.

Stabilizing oil prices, weaker-than-expected nonfarm payroll data, and downward revisions to Federal Reserve rate hike expectations have all failed to reverse the rise in global long-term bond yields. The reason behind this is persistently strong private-sector investment and financing demand. Driven by trillion-dollar investment, North America has become the first to break away from the "abnormal state" of low growth and low interest rates that followed the financial crisis. The global high interest rate environment, before the turning point of the AI investment cycle, is a normal state we must cope with.

The only demand insensitive to overseas high interest rates is North American AI and areas related to China's central fiscal expansion. The going-global and resources sectors, which delivered excellent holding experiences over the past few years, are both under pressure. In a weak demand environment, the scarcity of supply-cleared varieties stands out, and the anti-involution process still deserves attention next year. In terms of allocation strategy, in the short term, we can only use cyclical strength varieties and supply-cleared varieties to cope with the high interest rate environment, and we recommend closely tracking the turning point of the AI investment cycle.

Stabilizing oil prices, weaker-than-expected nonfarm data, and downward revisions to Fed rate hike expectations have all failed to reverse the rise in global long-term bond yields

Compared with the oil price shock phase in the first half of the year, current crude oil prices are closer to fluctuating at high levels. The implied inflation rate from the US 10-year TIPS has basically remained around 2.35% since early September, with the current reading at 2.33%. Recent oil price fluctuations have not yet pushed long-term inflation expectations further up, and inflation expectations are not the main factor behind this round of long-end rate increases. US September nonfarm payrolls added only 29,000 jobs, below the expected 90,000. Average hourly earnings rose just 0.1% month-on-month, below August's 0.3%, with year-on-year growth slowing to 3.0%, continuing to hit a new low since 2020. A series of weak employment and inflation data led the market to significantly revise down Fed rate hike expectations. The probability of a Fed rate hike in October implied by rate futures dropped sharply from 64.2% on September 25 to 17.7% on October 8. But these factors did not translate into weaker long-end government bond yields. Recently, 10-year government bond yields in the US, Japan, and the UK touched highs not seen since 2002, 1996, and 2007 respectively, while Germany and Australia rose to highs since 2011. The US dollar index continued to strengthen, rising from 99.7 on September 1 to 102.1 on October 8, breaking above the upper edge of the trading range since late April 2025. The largest recent decline in long-term bond yields occurred when OpenAI's ARR fell significantly below market expectations (early morning Beijing time on October 9), triggering market concerns about the sustainability of future AI capital expenditure. This indicates that the most important element in current long-term bond yield pricing is still expectations for the AI investment cycle, rather than short-term inflation and monetary policy direction.

Driven by trillion-dollar investment, North America has become the first to break away from the "abnormal state" of low growth and low interest rates that followed the financial crisis

1) Over the next year, high long-end bond yields may be the norm. US non-financial corporate net financing from the second half of 2025 to the first half of 2026 will reach $825.521 billion, an increase of 16.1% compared with the full year of 2025. Meanwhile, corporate net purchases of government bonds will fall from $107.878 billion in 2024 to $12.969 billion in 2025, and will even turn to net selling from the second half of 2025 to the first half of 2026. The private sector is both intensifying competition on the funding demand side and reducing its absorption of government bonds on the demand side. This is the essential reason why long-term bond yields remain at high levels. As long as AI investment and corporate financing demand do not cool substantially, the pressure on long-term funding supply and demand will be difficult to significantly ease. If the year-on-year growth portion of AI Capex in 2027 (market consensus of over $400 billion) needs to rely entirely on private-sector debt financing, then government bond yields will rise further. But high interest rates do not simply signal a major stock market correction. In particular, one should not directly compare the stock market's "E/P" with the 10-year government bond yield to judge the relative attractiveness of stocks and bonds. The difference between the S&P 500 earnings yield and the 10-year US Treasury yield was called the "FED Model" in the 1990s and was often used to measure relative stock-bond attractiveness. Empirically, a large body of research has proven that this spread has no predictive power for stock market returns. The predictive power of "E/P minus 10-year rate" is far inferior to that of "E/P" itself. In other words, historically, stock correction pressure has come from bubbles in their own valuations, rather than from capital crowding-out caused by stock-bond comparisons.

2) The core of predicting long-end rate direction is closely tracking the turning point of this round of the AI investment cycle. The cloud business profit margins of core CSPs and computing power rental prices are the forward-looking indicators. High interest rates do not mean the equity market has lost room to rise. If AI commercialization still fails to achieve greater breakthroughs, or the process of connecting with the physical world falls short of expectations, a weakening AI investment cycle will deal a significant blow to the equity market, mainly reflected in substantial downward revisions to earnings expectations. Conversely, as long as the intensity of AI investment continues, elevated long-end rates should be regarded as a normal state to adapt to for some time, rather than a reason to be消极 bearish on equity assets. However, if massive AI investment repeatedly fails to deliver a leap in productivity or corresponding incremental commercial space, this state will be difficult to sustain long term. At that point, the end of the AI investment cycle may be accompanied by a new round of global economic adjustment and collective clearing of risk assets. Companies will reduce investment, financing demand will decline accordingly, and demand for government bonds and other global safe assets will rebound, thereby driving long-end rates lower. The most effective forward-looking indicators for predicting the turning point of the AI investment cycle should be the cloud business profit margins of core CSPs and computing power rental prices, which represent the balance of real-world computing power supply and demand. These are the true risk indicators, while long-term bond yields are more of a result of investment and financing behavior.

The only demand insensitive to overseas high interest rates is North American AI and China's central fiscal expansion, while going-global and resources are both under pressure

1) North American AI and China's "Six Networks" construction are among the few directions that can withstand high interest rates in the future. In the current high interest rate environment, the directions globally that can still bear such rates and still grow are almost exclusively the trillion-dollar-level real investment in North American AI infrastructure. If the AI infrastructure investment cycle begins to slow, it will be extremely difficult for growth in other global areas to offset the downward pressure. Therefore, from the perspective of global capital markets, in an extremely high interest rate environment, the sector that may continue to rise into the late bull market and only correct at the end may be only AI. But looking at China's capital market separately, the situation is somewhat special. China's ample savings and capital keep it in the position of a global interest rate lowland. Private-sector investment and borrowing willingness are relatively weak, but the space and flexibility for central fiscal expansion are enormous, and it can better utilize the current abundant capital dividend. Considering policy tone and fiscal expansion philosophy, among domestic-demand-oriented directions that rely on debt financing, the new infrastructure represented by the "Six Networks" currently appears the most feasible. We estimate that during the "15th Five-Year Plan" period, the "Six Networks" are expected to drive a total upstream and downstream investment scale of RMB 53-72 trillion. In terms of pace, it will accelerate in 2027, and 2028 may be the peak during the period, corresponding to the peak revenue realization for listed companies in 2027-2028. The prosperity of general equipment, electrical equipment, and computing power ICT hardware may be relatively high overall.

2) The going-global and resources sectors, which delivered excellent holding experiences over the past few years, may come under pressure in a persistently high interest rate environment in the future. The going-global sector will face the potential squeeze on non-AI demand from a high interest rate environment (see "A-Share Strategy Focus 20260913 — Waiting for the Rate Hike to Land"), geopolitical issues may create additional trade friction costs, and a strong renminbi may generate sustained exchange losses. Recent China-EU economic and trade consultations have made some better-than-expected progress, at least avoiding direct trade friction on the hybrid vehicle issue. But over the weekend, the US launched another Section 337 investigation. Similar events usually do not directly damage companies' current revenue and profits, but they constantly suppress valuations. The problem for the resources sector is more direct: high real interest rates directly suppress the financial attribute premium of resource products, and the overall rise in industrial costs to some extent also restrains demand for upstream resource products from non-AI industries. With ROE at high levels and Chinese companies' willingness to expand production and invest overseas rising, even a quarterly-level commodity price adjustment may bring significant drawdowns to resource stocks and lower the holding experience. And holding experience is precisely one of the greatest advantages of resource stocks in the past.

A weak demand environment highlights the scarcity of supply-cleared varieties, and the anti-involution process still deserves attention

China's 10-year government bond yield fell from 3.63% in early 2015 to 1.69% on October 8, 2026. The weighted average corporate loan rate also fell from 4.64% in early 2020 to the latest 3.04%. At the same time, in the first eight months of 2026, private investment and manufacturing investment fell by 10.1% and 2.3% year-on-year respectively. The decline in financing costs has not yet brought about a broad return of investment expansion. This means companies' operating focus remains biased toward stock adjustment, more inclined to preserve cash flow and digest existing debt, rather than expanding investment through new leverage. Therefore, more attention should be paid to industries where capital expenditure continues to contract while profit margins gradually recover. Supply constraints in some industries also make the competitive position of advantaged companies more prominent. Therefore, anti-involution still deserves attention, but the allocation focus should shift from policy expectations to financial realization of improved competitive landscape. Data from the 2026 interim reports show that among 79 CITIC third-tier non-financial industries with total market capitalization of no less than RMB 300 billion, 24 industries simultaneously showed capital expenditure contraction and gross margin repair, of which 14 maintained this characteristic for at least three consecutive quarters. Although capital expenditure contraction cannot yet be directly equated with permanent capacity exit, continued investment restraint and margin recovery have provided important clues for identifying directions with improved supply landscape and strengthened competitive advantages. In manufacturing, anti-involution in chemicals, photovoltaics, lithium batteries, automobiles and components has entered deep water and the verification stage, with next year being the most important window. In services and distribution, platform economy, chain retail, express logistics, and hotels are also worth close tracking.

Use cyclical strength and supply clearance to cope with high interest rates, and closely track the turning point of the AI investment cycle

We remain optimistic about the market offensive window around the third-quarter earnings season. High interest rates will not hinder or affect the realization of this judgment. North American AI investment has strong demand support and financing capacity, making it the most resilient link in this round of global prosperity. A sustainable market offensive cycle cannot do without AI participation. However, considering that the pace of AI commercialization is still slow relative to the enormous investment scale, capital scarcity may constrain the ceiling of annual AI capital expenditure after 2028, leading to suppressed valuations across the AI hardware sector. Therefore, future offensive opportunities in AI may be more concentrated in new technologies and new themes supported by the logic of "product iteration under the same capital expenditure," while traditional institutional heavyweights may have more structural repair opportunities during earnings season. Before the AI investment cycle reaches a turning point (with key observation of CSP cloud business margins and computing power rental price trends), we expect such market moves to play out repeatedly. In terms of allocation strategy, focus on North American AI, China's "Six Networks" construction, and supply-cleared varieties driven by anti-involution. The AI plus energy and chemicals structure remains applicable in the current high interest rate environment. Within the technology sector, pay attention to optical communications, PCB, MLCC, gas turbines, wafer manufacturing, and especially value new technologies and new themes. For non-technology sectors, increase allocation to energy and chemicals, innovative drugs, dividend plays (coal, banks), and leading brokerages with going-global potential.

Risk factors

Intensified friction among China, the US and other parties in technology, trade, and finance; domestic policy intensity, implementation effects, or economic recovery falling short of expectations; domestic and overseas macro liquidity tightening more than expected; further escalation of conflicts in Russia-Ukraine, the Middle East, and other regions; China's real estate inventory digestion falling short of expectations.