CITIC SEC: The Federal Reserve's interest rate hike in September is a non-decisive variable, and the A-share market will remain dominated by fluctuations.

CITIC SEC released a research report stating that whether the Federal Reserve will raise interest rates in September is not enough to determine the direction of long-term interest rates and the equity market. In fact, recent price signals from the equity market indicate that the market has not fully priced in the expectations of interest rate hikes. High interest rates and the expectation of hikes have little impact on AI investments, but a potentially greater negative effect on non-AI sectors, which theoretically could exacerbate K-shaped divergence. However, the recent stock price facts are just the opposite, especially in the A-share market, where the technology sector has performed significantly worse than non-tech sectors. Overall, the A-share market is currently dominated by fluctuations; there is no need to panic about overseas interest rate issues, nor should one be overly aggressive just because the market's reaction to interest rates is stronger than expected. The deeper reason behind the domestic and foreign long-term interest rate spread is a misalignment in capital supply and demand. When "commodities going abroad" encounters more potential friction and disturbances, breaking the deadlock may depend on "finance going abroad." Whether the Federal Reserve will raise interest rates in September is not sufficient to decide the direction of long-term interest rates and the equity market. Currently, the rise in long-term U.S. Treasury yields is mainly driven by real interest rates, with limited contributions from inflation expectations. Since the beginning of this year, the nominal yield on 10-year U.S. Treasuries has risen by about 60 basis points, while the real yield has increased by approximately 50 basis points, accounting for over 83% of the contribution. At the same time, the correlation between long-term U.S. Treasury yields and the expectation of interest rate hikes is not stable; from mid-May to mid-July and from late August onwards, this positive correlation has weakened, with some periods turning negative. In other words, the rise in U.S. Treasury yields is not solely due to expectations of Federal Reserve rate hikes. The demand for financing in the private sector in the United States has increased, and risk-adjusted returns have outperformed public sector bonds, intensifying competition for funds between the public and private sectors, which is the main reason for the continuous rise in U.S. Treasury yields. As of July 2026, the amount of U.S. corporate bonds issued in the past 12 months has reached $2.4 trillion, a year-on-year increase of 25.1%. The proportion of this issuance to total debt issuance rose from 15.40% at the beginning of 2023 to 19.68%, indicating a continuous increase in the weight of private sector financing demand in the bond market. CITIC SEC believes that the expansion of AI infrastructure will further strengthen this trend. As long as the return rate of cloud infrastructure by cloud service providers (CSPs) does not decline, debt financing tools related to AI infrastructure will crowd out demand for government bonds, pushing long-term Treasury yields higher. Therefore, either long-term rates will continue to rise, possibly suppressing valuations in non-cyclical sectors, or the supply-demand relationship for computing power will relax, lowering cloud infrastructure returns, which could lead to a decrease in rates while lowering fundamental expectations (possibly a worse scenario). In either case, the difficulty of investing in the equity market is clearly increasing. Thus, whether or not the Federal Reserve raises rates in September is not the central contradiction determining the direction of long-term interest rates and the equity market; the key factor impacting the mid-term trend of long-term interest rates and the equity market is the capacity release cycle of AI infrastructure. High interest rates and the expectation of rate hikes have little impact on AI investments, but the technology sector has recently performed the weakest. 1) From a fundamental perspective, the investment in AI computing power is least sensitive to the Federal Reserve's interest rate decisions and U.S. Treasury yields. The profitability of CSP cloud businesses in North America is still in a stable expansion phase. The financial reports for Q2 2026 show that the operating profit margins of AWS, Google Cloud, and Intelligent Cloud reached 39.4%, 35.6%, and 40.6%, respectively, continuing to rise quarter over quarter. Large CSPs show low sensitivity to issuing debt in a high-interest-rate environment. According to S&P and Moody's rating data, Microsoft is currently rated AAA/Aaa, Alphabet AA+/Aa2, Amazon AA/A1, and Meta AA-/Aa3, with their interest-bearing debt/LTM EBITDA ratios ranging from 0.55 to 0.89, below the usual downgrade threshold of 1.0-1.5 times set by the rating agencies. Moreover, according to LSEG statistics, the five major tech companies have issued approximately $223 billion in new bonds since 2026, surpassing the roughly $109 billion total for 2025 and far exceeding the average annual level of about $28 billion from 2020 to 2024. Although bond issuance has significantly increased, the yield spread of 2-4 year dollar bonds relative to Treasuries remains narrow. As of September 3, the median Z-spread for the senior unsecured bonds of Alphabet, Amazon, and Meta with a remaining maturity of 2-4 years is only 34-36 basis points (with a median OAS of 51-53 basis points), only slightly widening from about 30 basis points in 2025. This is more of a technical change due to increased supply rather than a deterioration in credit quality. Moody's emphasized in its July report that Microsoft, Alphabet, Amazon, and Meta have some of the strongest balance sheets in the world. This is consistent with every historical supercycle driven by massive investments; whether during China's 2006-2007 or the U.S. 2004-2006 real estate financial cycle, demand was not damaged in the early stages of interest rate hikes. As long as there is still a gap in computing power, the EBIT margin of computing infrastructure can be maintained, and the marginal impact of rising financing costs on the income statement is very limited. 2) From the perspective of market pricing, the recent weakness in tech stocks does not reflect current credit issues of major firms, but rather long-term narrative and pricing issues. "Can the advantage in computing power convert into barriers to technological monopolies" is the most significant factor affecting the pricing of current tech stocks, determining whether AI capital expenditures continue to show "FOMO-style expansion" or revert to traditional public infrastructure models. The core premise supporting the narrative of computing power investment has been the progress in model capabilities, but currently, while the capabilities of models are improving, the gap between closed-source and open-source models is widening (and is difficult to catch up). CITIC SEC previously mentioned that RSI (Recursive Self-Improvement) and anti-distillation could be two significant factors. If the story of AI training AI can be articulated, it would mean that a substantial increment in computing power demand comes from AI agents themselves, and if anti-distillation can be articulated, it indicates that latecomers find it hard to catch up with front-line models at low costs. This week saw two events: one was OpenAI's release of GPT-6 Astra, beginning to tell the story of RSI being preliminarily realized; the other was Anthropic's release of Claude Fable 5.1, which clearly introduced anti-distillation mechanisms. CITIC SEC cannot yet determine whether these two model products will immediately change the market narrative, as measuring the effects of RSI or anti-distillation is not as easily perceivable by the market and the general public as Coding Agents or OpenClaw, and there is currently a lack of direct and explicit metrics for their effects. Both may help accelerate model iterations and widen the gap between models, but whether they can bring a magnitude of increase in computing power demand like agents remains uncertain. Before more evidence emerges, the market will continue to swing between pricing scenarios of "infrastructuralization of computing power" and "computing power as a barrier," with interest rates being merely a short-term disturbance in this process, not the main contradiction. Expectations of rate hikes theoretically exacerbate K-shaped divergence in non-AI sectors, but recent stock price trends are just the opposite. Although rate hikes will uniformly increase the financing costs across the entire economy, the discrepancy in prosperity across different industries leads to different slopes in the price-demand curve. Demand in the relatively weak non-AI sectors suffers more. As Federal Reserve Chairman Waller mentioned at the Jackson Hole summit, more than half of this year's capital expenditures are contributed by AI-related industries, credit spreads are at historically low levels, and overall financial conditions are not restrictive. However, some interest-sensitive non-AI sectors (such as agriculture and real estate) are already starting to feel pressure. The equity market has demonstrated a similar logic; heightened expectations of rate hikes tend to exacerbate K-shaped divergence between AI and non-AI sectors, particularly evident in the second quarter of this year. The core AI stock pools in China, the U.S., Japan, and South Korea have substantially outperformed the core non-AI stock pools, in sync with market expectations of the Fed's policy rate trends for December of this year. However, in the two weeks surrounding the Jackson Hole summit, global market K-shaped divergence did not continue to widen, despite the market's renewed expectation of the Federal Reserve raising interest rates during this period. More notably, on Monday this week, when the Chinese market opened, the declines in gold and precious metals sectors were close to those in U.S. stocks last Friday, with the declines over the first two days of the week largely in line with U.S. markets but rebounding distinctly on Wednesday, outperforming U.S. stocks, and showing deep V patterns during both Monday and Wednesday trading sessions. The trends in precious metals reflect that stock investors seem not to really believe that the Federal Reserve will maintain a tightening stance; the market believes either that the Fed will not raise rates in September or that stock prices have already reflected the tightening expectations. In fact, the stock market has not aggressively priced in expectations of the Federal Reserve raising interest rates. The global market's K-shaped divergence has not continued to strengthen; assets sensitive to Federal Reserve rates (such as precious metals stocks) have not shown significant adjustments, and the implied rate hike expectations from interest rate futures have not reached the strengths observed in May and June. Only long-term bond yields have reached new highs (with breakeven inflation expectations decreasing slightly). Examining the movements of these asset prices suggests that this round of rising long-term bond yields seems more closely related to investors downgrading their assessments of the intrinsic value of government bonds in developed countries (the crowding-out effect of AI investments on long-term government bond demand and a lack of trust in fiscal policies) rather than the influence of Federal Reserve rate policies. CITIC SEC believes that investors do not need to be overly anxious about the global sell-off of government bonds; this might merely represent the beginning of a global era of capital scarcity and the end of the low-interest-rate era. In the context of rapid advancements in AI technology, the declining demand for government bonds as traditionally perceived "safe assets" should be a long-term trend. The phenomenon of selling off U.S. Treasury bonds is a result of the logic of economic and market operations, and it should not be considered a reason forecasing stock movements; even if there is a short-term correlation between the two, it likely stems mainly from liquidity and sentiment factors. As for potential short-term risks in the stock market, the main concern is that investors may not have fully priced in the expectations of tightening. Recent strong performance in precious metals stocks indicates that investors are skeptical about the Fed raising rates. Based on this, if the Fed does raise rates in September and the market reacts by pulling back to address potential emotional shocks, it could subsequently downplay the impact of the Fed's rate policies and U.S. long-term yields, returning focus to fundamental factors. Conversely, if the Fed fails to raise rates, the market might not have sufficient upside probabilities or certainty. Overall, CITIC SEC believes the market remains mainly characterized by fluctuations, with no need to panic about overseas interest rates and no need to become overly aggressive due to a stronger-than-expected market reaction to rates. The deeper reason behind the domestic and foreign interest rate spread is a misalignment in capital supply and demand. To break the deadlock, one must observe the shift from "commodities going abroad" to "finance going abroad." Apart from China, most major economies have long been in a state of insufficient savings. This status was reasonable during the low-growth phase preceding the AI transformation, but the AI technology revolution has altered this condition. Even if the Federal Reserve turns dovish, it cannot change this status. The variable that can truly break this deadlock comes from China's potential for "capital going abroad." China's excess savings flowing into global markets could alleviate the capital supply-demand gap brought about by AI investments, lowering long-term rates and supporting equity market valuations; at the same time, China's capital can achieve higher expected returns, addressing the issue of "asset scarcity." Of course, to ensure that this portion of the "savings dividend" yields benefits for both domestic and foreign interest rate spreads, the government needs to utilize effective regulatory and tax collection measures so that overseas investment gains can improve domestic fiscal revenues, enhance redistribution efforts, and ultimately diffuse into consumption, changing the status of low rates, low capital returns, and low inflation. Since the beginning of this year, CITIC SEC has observed similar signs, with the government strengthening overseas tax collection and tightening certain non-compliant and uncontrolled foreign investment channels. Viewing these measures merely as contractionary policies for increasing tax sources may be somewhat biased. CITIC SEC believes these initiatives aim to loosen domestic capital going abroad in a more compliant and monitorable manner in the future, using part of the difference in returns between internal and external capital to replenish the domestic economy, improving fiscal conditions, subsidizing domestic demand, and forming a new cycle. This is also an essential path for enhancing China's influence in the global financial arena in the long term. From this perspective, if the essence of the bull market in recent years has been "commodities going abroad" (whether AI or non-AI), then amidst increasingly complex trade forms, the next wave of mid-term market trends will likely be accompanied by "Chinese capital going abroad" and "financial capital going abroad," with the initiation of financial stocks, especially non-bank financials, serving as a significant signal.