CITIC SEC: Rate hikes difficult to shake AI valuation, non-AI weakness needs to break through its own narrative.

CITIC Securities released a research report stating that the market driven by AI this year is driven by massive infrastructure investment, resulting in a "bottleneck trade" similar to the bull market driven by investment and heavy asset companies in 2006-07, rather than the internet bubble market. It is difficult for interest rate hikes to affect the valuation of "AI cyclical stocks" unless interest rate hikes truly impact the terminal demand for AI, commercialization assumptions, and capital expenditure growth rate. The process of interest rate hikes globally first affects sectors with relatively weak demand growth, and the K-shaped differentiation between AI and non-AI is established globally. However, due to the simultaneous return of the strong dollar narrative and the overall market presenting a structural adjustment of stock funds, A-share non-AI cyclical sectors are significantly weaker compared to overseas benchmark companies. Similarly, with K-shaped differentiation, the breadth of the A-share market is relatively insufficient compared to overseas. Changing the weakness of non-AI sectors will require positive changes in their own narratives in the future, or changes in the funding situation, rather than waiting for AI adjustments. The main points of CITIC Securities are as follows: If the current AI-driven market is likened to the bull market in China in 2006-07, then "interest rate hikes" will not be the turning point for the end of the market 1) The current market is driven by massive investment, more like the bull market in 2006-07, rather than the internet bubble market. The global AI market has been clearly interpreted as a "bottleneck trade" this year. U.S. hardware storage, optical communications, and semiconductor equipment leaders have performed well, while traditional platform companies and AI application leaders have generally performed poorly. Compared to the peak in 2000, the forward PE ratios of the Nasdaq 100 are as high as 60 times, compared to only 25 times currently; while the average rolling PE ratio for the "Big Four" in 2000 soared to 82.7 times, the current "Seven Giants" are at 38 times. The Asia-Pacific region's market is mainly driven by "supply and demand" rather than just valuation. The Asia-Pacific core index has outperformed U.S. stocks this year, especially the KOSPI50 (+160%) and the Nikkei 225 (42%), which are deeply embedded in the AI supply chain, with corresponding PE-TTM ratios of only 10 times and 22 times, respectively. Similar characteristics are also present in the domestic market. At the macro level, the overflow of AI infrastructure investment and the upgrading of China's technology industry are very similar to the global industrial demand shifting eastward in 2006-07 and the explosive growth of China's foreign trade exports; at the market level, sectors such as capacitors, glass fibers, copper foils, IGBTs, silicon carbide, silicon wafers, etc., which have shown good returns this year, were traditionally seen as typical cyclical stocks two years ago within the traditional supply-demand framework. The PBs of industry leaders are generally between 1-3 times, but driven by the dividend of North American AI Capex and the narrative of domestic substitution, these sectors have undergone a deep restructuring of valuations, turning into PEG valuations or even PS valuations. 2) Interest rate hikes are difficult to impact the valuation of "AI cyclical stocks" unless interest rate hikes truly affect AI terminal demand, commercialization assumptions, and capital expenditure growth rates. This week, Powell, as the chairman of the Federal Reserve, made his first appearance, breaking the forward guidance mechanism introduced by Bernanke in 2012, refusing to submit personal interest rate forecasts, emphasizing a focus on inflation and employment to restore the credibility of the Federal Reserve, and the market currently expects and prices in a 25 basis point rate hike in October without a rate cut this year. However, the path of impact of interest rate hikes on growth stock markets driven by valuation and cyclical stock markets driven by supply-demand gaps is quite different. The Federal Reserve began raising interest rates in June 1999, and 7 months later the Nasdaq index peaked and entered a bear market; whereas for the bull market from 2004-2007, the Federal Reserve first raised interest rates in June 2004, and the bull market ended 38 months later in October 2007. For this massive investment-driven "bottleneck trade," interest rate hikes are unlikely to directly impact the valuation of "AI cyclical stocks" unless they truly affect terminal demand and capital expenditure growth rates, or the competitive landscape, and the narrative of AI realization faces significant challenges. Therefore, interest rate hikes (especially in the early stages) are unlikely to directly impact the valuation of "AI cyclical stocks". The global process of interest rate hikes has affected sectors with relatively weak demand growth, and the K-shaped differentiation between AI and non-AI is established globally Expectations for loose policies starting from 2024 have shifted in the past six months, with the US, Europe, and the UK transitioning from a rate cut cycle to a neutral one. The process of interest rate hikes has first affected sectors with relatively weak demand growth, and the K-shaped differentiation between AI and non-AI is established globally. AI has been almost the only main trend in markets around the world this year, with AI consistently outperforming non-AI globally, but the nature of differentiation varies greatly in different countries. In terms of performance, the cumulative excess returns of AI relative to non-AI from the beginning of the year are: South Korea +157%, China +143%, Japan +134%, and the United States +96%, with greater differentiation in Asia than in the United States. However, the AI/non-AI market PE ratio calculated based on the overall market value divided by the overall earnings is only 1.48 times in the US, 1.45 times in Japan, and 1.27 times in South Korea, while it is as high as 9.8 times in A-shares (with an AI basket of around 92 times PE and non-AI only around 9 times), a full magnitude higher than developed markets. In other words, South Korea and Japan have "rapid profit growth, with low valuations" (at the beginning of the year, the PE of AI was even lower than non-AI), indicating profit-driven and relatively healthy conditions, the United States has relatively moderate profit and valuation premiums, while the rise in A-share valuations is significantly higher. The return of the strong dollar narrative, A-share non-AI cyclical sectors are significantly weaker compared to overseas Judging from Powell's statements at the June FOMC meeting, the Federal Reserve emphasizes restoring credibility, focusing on the targets of inflation and employment. On the economic industrial side, based on the latest guidance from various companies' financial reports, the capital expenditure scale in the field of artificial intelligence by some large technology companies in 2026 is close to $700 billion, most of which are initiated and implemented in North America by the private sector, establishing a monopoly competitive advantage compared to the rest of the world. In this combination, the dollar's strength has returned, with the narrative of the "Great Moderation" of the 1990s making a comeback. This can be seen from the recent divergence between the dollar index and long-term U.S. bond yields, the 10-year U.S. bond yield has dropped from 4.6% a month ago (May 18) to around 4.45% on June 18, while the dollar index has risen from 99.0 to 100.8 during the same period, indicating that the recent strength of the dollar cannot be explained solely by the initial shortage of dollar liquidity due to the early stages of the Middle East conflict. This partly explains why A-share non-AI cyclical sectors have been weaker compared to overseas since May. In fact, for many globally priced industrial goods, global infrastructure demand driven by AI is still growing, China's "Six Networks" plan is steadily advancing, and global defense spending has entered a new phase of expansion, with traditional resource countries in the Middle East, Latin America, and Southeast Asia increasing industrial investments, indicating that there is no narrative for a downturn in demand. The performance outlook for many industrial manufacturing sectors on the A-share market this year has generally been revised upwards, which is completely different from the continued downward revisions in the consumption sector, but their stock performance has converged and is clearly weaker than overseas, a similar situation has also occurred in the brokerage sector. Factors such as pre-emptive negative narratives and liquidity issues may be at play, the continued net redemptions of wide-based ETFs have continued to suppress non-AI sectors lacking clear catalysts, and during the index adjustment stage (May 28 to June 12), the net redemptions of wide-based ETFs slowed significantly, and related sectors saw a brief recovery; however, with the rebound in the index, this week (June 15 to June 18), the 2025 annual report showed that the China Investment Corporation continues to net sell 4 Shanghai and Shenzhen 300 ETFs, with a total net outflow of 42.2 billion yuan. The strong dollar narrative and net redemptions of ETFs are key factors currently suppressing non-AI cyclical sectors. Changing the weakness of non-AI cyclical sectors still requires positive changes in their own narratives in the future, rather than waiting for AI adjustments This change could come from a sharp decline in oil prices following the opening of the strait for navigation, which lowers inflation expectations. Alternatively, it could come from simultaneous recovery in global non-AI industrial production and social activities (such as Asian countries resuming production and increasing supply as oil and gas prices fall). After all, the global environment we are in cannot be described as a recession, and it is expected that industrial production will not weaken in the process of AI development, national defense, and accelerated industrialization in developing countries, it just needs more evidence for the market to see. In fact, discussions about silicon-based and carbon-based differentiation is more common domestically in China, which may be due to weak consumption growth data globally. China's fixed asset investment is weak, but it is tilted towards advanced technology manufacturing, which is a good thing in the long term, slowing down the rate of capital accumulation in traditional industries reduces internal competition, and emerging manufacturing industries continue to increase investment to improve competitiveness, thus showing greater profit potential for technology and manufacturing sectors in A-shares. In terms of allocation, we still recommend adhering to the AI+energy chemical structure. In the AI sector, we continue to favor species with relatively low crowdedness, such as storage, gas turbines, diesel generators, and some semiconductor equipment and materials. In the energy chemical sector, we are bullish on the performance realization of electrolytes and additives, as well as membranes; in the chemical sector, the decline in oil prices and volatility will bring about additional stocking needs, and the peak in macro liquidity expectations will be a potential turning point in the future, we currently prefer species with significant potential for cost reduction, strong demand, and low valuations, such as refrigerants, phosphor-chemicals, spandex, dyes, and large refineries; in the non-ferrous sector, we recommend metals with some AI exposure but are temporarily suppressed in valuation due to the interest rate narrative, such as tin, copper, and some small AI metals (tungsten). In addition, we continue to recommend increasing allocation to undervalued brokerages, as liquidity constraints and other issues may gradually dissipate in the second half of the year, with interim reports also acting as catalysts. Risk factors Intensification of friction in the fields of technology, trade, and finance between China and the United States; domestic policy efforts, implementation effects or economic recovery falling short of expectations; unexpected tightening of macro liquidity at home and abroad; further escalation of conflicts in areas such as Russia, Ukraine, and the Middle East; China's real estate inventory digestion falls short of expectations.