The AI industry trend has not reversed; the recent correction stems more from funding and sentiment pressures rather than a fundamental deterioration in earnings trends.
Zhitong Finance APP reports that CICC released a research note stating that global assets face multiple challenges in September. AI-related bond issuance may accelerate, increasing uncertainty in U.S. Treasury yields; Waller’s hawkish shift has heightened the risk of Federal Reserve rate hikes; escalating U.S.-Iran tensions have led to a notable rebound in oil prices. The firm advises maintaining confidence and patience, as the timing for policy correction may be approaching. Short-term volatility will not reverse the global trend of loose liquidity; investors should utilize market fluctuations to increase allocations to Chinese and U.S. technology stocks and gold at lower prices.
CICC’s key views are as follows:
September may witness a new wave of AI financing shocks
In mid-August, long-end U.S. Treasuries underwent a rapid adjustment, with the 30-year U.S. Treasury yield briefly surpassing 5.3%. The firm previously pointed out that the primary driver behind this rise in long-end rates was not an increase in long-term Treasury supply by the U.S. Department of the Treasury, but rather an expansion in the supply of AI-related corporate bonds. As tech giants continue to expand capital expenditure and extend financing durations, long-duration AI corporate bonds are competing with U.S. Treasuries for the same pool of duration-sensitive capital. The increased supply of new corporate debt has raised term premiums across the market, passively pushing up long-end U.S. Treasury yields.
For the full year, the firm expects the net supply of investment-grade corporate bonds in 2026 to approach $1 trillion, a year-on-year increase of over 70%, while the net supply of interest-bearing Treasury bonds is projected to decline to $1.2 trillion. This reflects that pressure in the bond market supply stems not from U.S. Treasuries, but from corporate bonds.
As the supply of AI-related bonds increases, credit spreads for leading U.S. technology companies have widened to varying degrees recently.
September is traditionally the peak season for investment-grade corporate bond issuance in the United States, and supply pressure in the U.S. credit bond market may rise again. Underwriters summarized by Bloomberg expect approximately $215 billion in investment-grade corporate bonds to be issued in September.
Against the backdrop of significant depletion of long-duration capital, a new round of concentrated issuances may once again raise financing costs and term premiums for corporate bonds, causing disruptions to long-end U.S. Treasuries. Meanwhile, the stock market also faces certain financing pressures. Year-to-date, U.S. IPOs have raised approximately $137.6 billion, a 464% year-on-year increase, with SpaceX alone raising about $85.7 billion in a single offering, marking the largest IPO in U.S. history.
AI-related equity financing continues to heat up in the autumn. Anthropic filed its IPO application in June and is currently preparing for listing, with market expectations suggesting it could launch its IPO as early as October. If AI financing is released in a concentrated manner, bond financing may increase duration supply and disrupt long-end interest rates, while IPOs may divert risk capital; both factors jointly exert pressure on high-valuation assets.
Geopolitical risks are rising, and higher oil prices are delaying progress in inflation improvement, although core inflation may remain at low levels.
Since late August, tensions in the Middle East have escalated again, with cargo traffic through the Strait of Hormuz falling to its lowest level since May. Renewed conflict between Saudi Arabia and Houthi forces has led to attacks on Saudi energy infrastructure, sparking market concerns over potential energy supply shortages. As a result, Brent crude oil prices have risen steadily from a阶段性 low of $87.8 per barrel on August 26, breaking above $100 per barrel during intraday trading on September 9.
The bank believes that rising oil prices may drive a temporary rebound in the U.S. headline CPI for August, but core inflation is expected to continue cooling. The U.S. August CPI data is scheduled for release on Friday, September 11. CICC's Asset Allocation Team forecasts that the U.S. headline CPI will rebound by 0.36% month-on-month (previous: 0.07%; consensus: 0.4%), while the year-on-year rate remains at 3.37%. Core CPI is projected to stay at a low monthly increase of 0.18% (previous: 0.22%; consensus: 0.2%), with the year-on-year rate declining to 2.35%.
The rebound in headline inflation is primarily driven by the energy component: although August is typically a seasonally weak month for oil prices, prices rose contrary to seasonal trends this year, pushing up headline CPI.
Core CPI is likely to remain subdued, mainly due to two factors. First, high-frequency data shows that wholesale used car prices have declined more sharply over the past two months; this transmission to retail prices may slow the month-on-month growth of the used car CPI component. Second, falling import prices and tariffs are also exerting downward pressure on inflation for other core goods.
The market widely views Friday's CPI data as decisive for the Federal Reserve's interest rate decision in September. However, if the above forecasts materialize, the CPI may not provide a clear signal ahead of the Fed meeting. If Chair Walsh intends to raise rates, he could emphasize the month-on-month rebound in headline inflation and the fact that year-on-year improvements have stalled. Following the logic of his Jackson Hole speech, as long as inflation does not fall "fast enough," rate hikes should proceed. Conversely, if Walsh prefers not to raise rates, he could highlight that core CPI rose by only 0.2% month-on-month and continues to decline slightly on a year-on-year basis.
The same data, viewed from different perspectives, can lead to diametrically opposite policy implications. Of course, considering statistical errors, the bank's forecasts may also prove incorrect. If inflation turns out significantly higher or lower than predicted, it may provide clearer policy signals before the September Fed meeting.
Fed Chair Walsh explicitly adopted a hawkish tone at Jackson Hole, raising the risk of a rate hike in September.
Walsh had previously argued that "if inflation remains persistently high, further rate hikes may be necessary." This time, he further stated that inflation must not only decline but do so "clearly and at a sufficiently fast pace" toward the 2% target; otherwise, the Fed must still take action. For the first time, Walsh set explicit requirements for the speed of disinflation, significantly lowering the threshold for a rate hike.
Unlike the rapid disinflation observed over the past two months, the recent rebound in oil prices was already poised to slow the pace of decline, thereby increasing the risk of a September rate hike. Another key takeaway from the Jackson Hole symposium was Walsh's refusal to interpret recent data improvements as a change in trend. Although the U.S. labor market has cooled compared to earlier periods, consumption growth shows signs of marginal slowdown, and CPI has cooled for two consecutive months, Walsh still emphasized that underlying inflation improvements remain limited. He attributed the labor market slowdown mainly to a contraction in labor supply, noting that the economy and job market remain resilient.
Walsh's forced hawkish interpretation of dovish data may be an attempt to restore the credibility of the U.S. dollar. The bank believes that U.S. economic fundamentals do not actually support a rate hike: the underlying center of gravity for U.S. inflation is not high, and the rise in oil prices has not generated significant "second-round effects." Inflation is still expected to continue moving toward 2%, and the labor market is cooling. Additionally, with the midterm elections approaching, further tightening faces political constraints.
Therefore, Waller’s recent hawkish remarks may have been merely rhetorical. The bank’s baseline scenario remains that there will be no rate hike in September. However, the possibility that the Federal Reserve might “overcorrect” by raising rates to restore its credibility can no longer be ruled out. Given the evident lowering of the threshold for rate hikes at the Jackson Hole symposium, the Fed faces a dilemma at its September meeting: failing to raise rates could undermine policy credibility, while raising them would further hurt the economy and increase political costs, resulting in heightened policy uncertainty.
The market has already priced in rate hikes by the European Central Bank and the Bank of Japan for September. If the Federal Reserve also initiates a rate hike, the simultaneous tightening by the Fed, ECB, and BOJ would create a phased shock to global liquidity conditions.
The timing for policy correction may be approaching, and the pressure on AI-related bond issuance could ease temporarily after September.
The bank believes that multiple U.S. policies are currently misaligned. If economic and market conditions continue to deteriorate, policy corrections could occur at any time:
First, with only two months remaining until the U.S. midterm elections, the optimal strategy for the United States is to end the conflict as soon as possible to reduce oil prices and inflationary pressures.
Second, U.S. employment, consumption, and inflation have all slowed over the past two months. The rebound in some employment and inflation data in September lacks sustainability. U.S. inflation is highly likely to fall to around 2% next year, and the AI revolution will lower the long-term inflation中枢 (central tendency). In fact, the Federal Reserve is well-positioned to wait for inflationary pressures to ease naturally; it could choose not to raise rates now and initiate rate cuts once inflation declines. Even if rates are raised in September, a quick policy correction could follow, restarting the rate-cutting cycle. Regardless of whether rates are raised in September, monetary policy is likely to revert to rate cuts in the future.
Third, financing pressure on AI-related bonds may marginally ease after September. September is traditionally the peak season for U.S. credit bond issuance. The bank expects issuance pressure to decline seasonally in October and November, potentially reducing disturbances to term premiums and long-end interest rates.
Increase allocations to Chinese and U.S. technology stocks and gold on dips, while U.S. Treasuries are ranked relatively lower. The September FOMC meeting is a critical juncture.
Based on the above analysis, the recent pullback may create opportunities to buy on dips, with the key factor being when the turning point will emerge.
If the FOMC raises rates at its meeting on September 16, it may not continue hiking thereafter. The market might begin to price in the exhaustion of negative factors, leading stocks and gold to initially fall and then rise after the meeting. If there is no rate hike in September, the short-term risk of a hike will be directly alleviated. However, the Fed’s previous strong hawkish rhetoric without subsequent action could further undermine policy credibility, which would be more favorable for gold.
From the perspective of event timing, the period following the September FOMC meeting may present a rebound window with a favorable risk-reward profile, which investors are advised to prioritize. Meanwhile, given uncertainties surrounding the U.S.-Iran situation and economic data, the market could also initiate a rebound prior to the FOMC meeting (for instance, if Trump rapidly resolves the conflict, if U.S. CPI falls significantly short of expectations, or if other major policy adjustments occur). Trading tactics should therefore remain flexible. As the bullish trends in gold and Chinese and U.S. technology stocks remain intact, the bank believes there is no need to mechanically wait for a specific point in time. If the market experiences a notable correction in the coming weeks, investors may consider gradually increasing their allocations to equities and gold before the Federal Reserve meeting. By asset class:
(1) Gold remains the clearest candidate for an overweight position. Whether the Federal Reserve ultimately pivots to easing to improve liquidity, or recent policy missteps undermine confidence in the U.S. dollar, the medium-term rationale for holding gold remains unchanged. Corrections merely provide opportunities to increase positions.
(2) Chinese and U.S. equities can still be accumulated on dips, particularly in the technology sector. The AI industry trend has not reversed; recent adjustments stem more from funding and sentiment pressures rather than a fundamental deterioration in earnings trends. If policy and liquidity pressures ease, high-valuation assets may exhibit greater rebound elasticity.
(3) There is no urgent need to bottom-fish U.S. Treasuries. Risks related to AI financing and interest rate hikes in September may continue to disrupt long-end yields, making long-term bonds less certain than gold and equities. Rather than betting on a rapid decline in long-end rates, the bank maintains a more neutral view on U.S. Treasuries, recommending patience as policy risks and supply pressures unwind.