CITIC Securities: Navigating High Interest Rates Through Sector Momentum and Supply-Side Clearing, Tracking the AI Investment Cycle Inflection Point
nashnova research
CITIC Securities argues that global long-bond yields have hit multi-decade highs driven not by inflation but by relentless private-sector borrowing demand; until the AI investment cycle reaches an inflection point, high rates are the new baseline markets must price in.
Why won't global long-bond yields come down?
Oil has stabilized, U.S. September non-farm payrolls missed expectations, and rate-hike bets have faded — yet 10-year government bond yields in the U.S., Japan, and the U.K. touched highs not seen since 2002, 1996, and 2007 respectively.
This means → the force pushing yields higher is structural, not episodic: private-sector investment and financing demand remains so strong that it is crowding out flows into sovereign bonds.
In plain terms = companies are borrowing aggressively to invest; that leaves less money chasing government debt, so yields can only go up.
What does the data say about corporates "competing with governments for capital"?
Net financing by U.S. non-financial corporates reached $825.5 billion in H2 2025–H1 2026, up 16.1% from full-year 2025.
Over the same period, corporate net purchases of U.S. Treasuries plunged from $107.9 billion in 2024 to just $13.0 billion in 2025 — and flipped to net selling from H2 2025 onward.
This means → corporates are simultaneously borrowing more and buying fewer Treasuries — a double squeeze that pins long-end yields at elevated levels.
Why has the AI investment outlook become the key variable in rate pricing?
The largest recent drop in long-bond yields coincided precisely with OpenAI's annualized recurring revenue (ARR — subscription revenue on an annualized basis) coming in well below market expectations.
This reflects a market whose real anxiety is not inflation but whether AI capex — the trillions tech giants spend on data centers and chips — can be sustained.
CITIC Securities argues the most effective leading indicators for the AI cycle inflection point are cloud-business profit margins at core cloud service providers (CSPs) and compute-rental pricing — real-world proxies for whether compute supply and demand are still expanding.
Which sectors can withstand high rates?
Globally, almost the only pocket that can absorb high rates and keep growing is North American AI infrastructure and its trillion-dollar-scale real spending. If AI buildout slows, other sectors cannot offset the drag.
China is a special case: abundant savings, the world's lowest rates, and substantial room for central-government fiscal expansion. The "Six Networks" new-infrastructure program (the infrastructure investment framework for the 15th Five-Year Plan period) could drive RMB 53–72 trillion in upstream and downstream investment, accelerating in 2027 and possibly peaking in 2028.
In plain terms = in a high-rate world only two engines can still spend big — America's AI buildout and China's new infrastructure. Everything else is being squeezed.
Are the good days over for the "going-global" and resources trades?
The going-global theme faces a triple headwind: high rates compressing non-AI demand, geopolitical friction raising trade costs, and a strong renminbi generating persistent FX translation losses.
Resource plays are equally pressured: high real rates suppress the financial-attribute premium, while rising industrial costs dampen non-AI industries' demand for upstream commodities.
This means → the two trades that delivered the best shareholder experience over the past few years are entering a phase where their underlying logic is being eroded.
How to pick the "anti-involution" winners — and where to allocate?
China's 10-year government bond yield has fallen from 3.63% in early 2015 to 1.69%; corporate lending rates have dropped from 4.64% to 3.04% — yet private and manufacturing investment still fell 10.1% and 2.3% year-on-year respectively. Cheaper borrowing has not revived broad-based investment.
Among 79 CITIC Level-3 non-financial sectors with market caps above RMB 300 billion, 24 show simultaneous capex contraction and gross-margin recovery; 14 have sustained this pattern for at least three consecutive quarters. This means → the allocation focus should shift from policy bets to earnings delivery driven by improving competitive structures.
Within tech: optical communications, PCB, MLCC (multilayer ceramic capacitors — among the most fundamental passive components in electronics), gas turbines, wafer fabrication, and emerging technology themes. Outside tech: add petrochemicals, innovative pharma, dividend plays (coal, banks), and leading brokerages with overseas-expansion potential.
市场有风险，内容仅供研究参考，不构成投资建议。