On October 16 during Beijing time midday trading hours, Fed Chair Kevin Warsh held a dialogue with Georgieva during the IMF and World Bank annual meetings in Bangkok, reiterating that the Federal Reserve will achieve price stability and describing the September rate hike as removing "a portion of easing."
Regarding the Fed's FOMC monetary policy meeting scheduled for October 27-28, he explicitly stated he would not prejudge the Fed's monetary asset path, nor would he prematurely announce possible actions at remaining meetings this year, emphasizing that each Fed decision will form policy judgments based on output, employment and inflation rather than a preset policy path, and reiterating his personal preference that the Fed provide no forward guidance whatsoever.
Notably, Warsh's series of remarks at the IMF reinforced the judgment that frontier AI technology will change the economy over the long term, yet he gave no commitment to pausing rate hikes in October. For global semiconductor sector stocks that have recently suffered violent selling and fallen into correction territory, the positive aspect is that he did not further strengthen near-term rate hike expectations; however, whether Treasury yields can continue to cool and whether the recently battered AI computing theme stocks and the broader semiconductor and tech-related sectors can launch a counterattack still depends on whether inflation, corporate financing conditions, and AI profit realization can improve simultaneously.
More noteworthy than "not prejudging monetary policy" is his statement that since the September FOMC meeting, economic evolution has generally met expectations, with structural growth and investment surges currently present. This means he has not equated the recent semiconductor-led global tech stock correction with a breakdown in the economic growth logic. At the same time, he emphasized that most of the world is tightening policy in sync, and all countries need to maintain price stability; US policy transmits overseas, and overseas policy also affects the US in reverse. His policy framework remains one of safeguarding inflation and institutional credibility while preserving room for judgment.
Before his remarks, MUFG's latest disclosure showed rate futures market pricing implied about a 27% probability of an October rate hike, but still fully priced in a 25 basis point hike before year-end — therefore both before and after Warsh's speech, rate futures markets were consistently trading the possibility that the timing of rate hikes might be pushed back, rather than the tightening cycle having ended. Wall Street giants have recently been converging further toward "continued rate hikes, extended tightening cycle" in their Fed predictions, with the latest explicit adjustments coming from Citi and UBS.
The Fed already hiked rates by 25 basis points in September, and this round of institutional revisions focuses on how much more will follow. On October 14, Citi withdrew its forecast of "no further rate adjustments this year," instead predicting another 25 basis point hike in December, followed by a pause until rate cuts begin around June 2027. UBS Global Wealth Management, between October 12 and 15, extended its previous "one more hike in December" path to 25 basis point hikes in both December this year and March next year; combined with the September action already implemented, the entire cycle totals 75 basis points of cumulative hikes, with rates then expected to be held near year-end 2027.
The core driver behind Wall Street's reassessment of rate hike predictions is that energy shocks are beginning to spread to broader prices, while the economy has not yet weakened enough to force the Fed to prioritize growth support. September core CPI rose 0.3% month-over-month, exceeding the 0.2% market expectation for two consecutive months, with the year-over-year rate rising from 2.4% to 2.6%; housing, airfares, automobiles, clothing and tech-related goods all rose together, making it harder for policymakers to view the inflation rebound as an isolated oil price disturbance. The subsequently released September PPI rose 1.3% month-over-month and 6.3% year-over-year, further reinforcing concerns about upstream cost pass-through to end prices.
AI first drives demand, then releases productivity: why the computing boom coexists with high rates. In his remarks, Warsh described AI as the most profound economic transformation of his lifetime, and the key to understanding this judgment is his simultaneous emphasis that AI's impact on the demand side will appear before supply-side improvements. From the perspective of data center engineering and the AI computing industry chain investment, companies building computing clusters need to first purchase GPUs, TPUs and ASICs, HBM, networking equipment, and invest in power, cooling, facilities and financing; while the cost reductions and efficiency gains formed after AI is widely embedded in R&D, production and corporate processes require deployment, organizational adjustment and commercial validation.
This time lag allows AI to simultaneously boost semiconductor orders while increasing capital and power demand, pushing up some prices and financing costs in the short term. Therefore, Warsh's AI optimism does not naturally correspond to falling rates. Productivity improvements help expand supply and suppress unit costs, but higher return on investment expectations may also attract more companies competing for capital, pushing up real interest rates. For the trajectory of 10-year and 30-year Treasury yields, the market must also simultaneously factor in future policy rates, fiscal financing and term risk compensation. Strong computing demand and elevated long-term capital costs could completely arise from the same round of AI investment expansion; semiconductor earnings growth and valuation pressure can therefore occur simultaneously.
His latest remarks also revealed how policy research is following this change: an AI field hearing is planned for November 16 in the San Francisco Bay Area, with a Fed working group forming new research insights to assist FOMC judgments; AI is also expected to improve the central bank's ability to understand data, though short-term employment impacts remain uncertain. For private technology investment, the Fed will not judge who is right or wrong, but will focus on financial tail risks that can be addressed through policy. These signs all mean that the Fed and other global central banks will increasingly value AI's contribution to growth, and are also examining financing structures and risk transmission, without promising protection for AI asset prices.
Treasuries have already cooled; a semiconductor counterattack still requires credit financing cost cooling and earnings resonance. Long-end yields have indeed begun to fall. According to unified US Treasury Department data, on October 15 the 10-year and 30-year Treasury yields were 5.10% and 5.43% respectively, both down 9 basis points on the day; for the week through the 15th they fell 14 and 17 basis points respectively, and compared with end-September they fell 19 and 21 basis points. However, both remain 35 and 18 basis points higher than end-August, indicating that the recent cooling only partially reversed the previous financing cost shock.
Notably, an important driver of this decline is safe-haven bond buying triggered by turbulence in AI stocks and credit markets. TD Securities believes investors are lowering growth expectations, and Treasuries are once again playing their role as an equity hedge tool. This creates a seemingly contradictory market combination: pressure on computing stocks pushes Treasury yields lower, while lower yields are temporarily insufficient to eliminate computing financing concerns. Corporate debt costs are composed of benchmark rates plus credit spreads; if the former falls while the latter widens, data center projects reliant on external financing may still face persistently high actual borrowing costs.
Global stock markets are also in a repair window after sharp declines as Warsh spoke. As of October 15, the Philadelphia Semiconductor Index fell 0.88%, accumulating a decline of about 12.4% over seven consecutive trading days, down about 7.9% for the week and about 8.3% for October, though still up about 63.5% year-to-date, with short-term momentum clearly damaged. On October 16 during Beijing time Asian midday trading hours, South Korea's benchmark KOSPI index — known as the "AI memory chip" bellwether — rebounded 3.17% in a single day, with Samsung and SK Hynix recovering in tandem; this rebound also preceded Warsh's remarks, though the rally is continuing.
Warsh's remarks this time provided some support for sentiment repair in the battered global semiconductor sector, but the combination truly capable of driving a sustained semiconductor counterattack is "long-term yield decline + credit spread stability + continued realization of computing orders and cash flow." If financing pressure eases while AI demand remains strong, the market will have conditions to pay higher valuations for growth again; among them, chip and equipment leaders with abundant cash flow, stable customer funding sources and strong order conversion capability are more likely to benefit first. The long-term AI transformation emphasized by Warsh provides a macro explanation for this growth theme, but whether the short-term rebound can escalate into a trend still depends on whether declining capital costs can truly transmit to industrial investment.
The "three major reinforcements" recently described by Wall Street financial giant Goldman Sachs (buybacks, retail investors, and foreign capital — these three forces) are expected to improve the funding environment for AI assets to regain pricing. Goldman Sachs' fund flow team pointed out in its latest report "Witching Hour" that according to "Witching Hour" estimates, about 48% of S&P 500 constituents will re-enter the buyback open window by end-October; historically, November contributes about 10.1% of full-year buyback execution volume, and combined with the fading of mutual fund fiscal year-end tax selling pressure, year-end retail fund inflows, and overseas allocation demand after the midterm elections, stock supply-demand relations may gradually improve. Among these, about $190 billion in overseas buying represents a potential scale extrapolated from historical patterns.
For semiconductors that have experienced concentrated position reduction, the restoration of buying means companies could trigger position replenishment even by merely proving that orders and earnings have not significantly deteriorated. However, tech sector buyback execution volume fell 40% year-over-year, indicating that capital expenditure is competing with shareholder returns for funds; for AI stocks to continue attracting incremental allocation, they still need to prove that capital investment related to GPUs, network infrastructure, and key AI infrastructure resources such as data center optical interconnects and CPUs can generate higher future cash flows.
Comments