In a series of remarks at the IMF, Wash reinforced the view that cutting-edge AI technologies will reshape the economy over the long term, while refraining from committing to a pause in rate hikes in October. For global semiconductor stocks, which have recently faced sharp sell-offs and entered a correction phase, the positive takeaway is that he did not further bolster expectations of near-term rate increases.
On October 16, during the midday session in Beijing time, Federal Reserve Chair Kevin Warsh engaged in a dialogue with Kristalina Georgieva on the sidelines of the IMF–World Bank Annual Meetings in Bangkok, reiterating that the Fed remains committed to price stability and characterizing the September rate hike as a step to unwind "some of the accommodative stance." Regarding the upcoming Federal Open Market Committee (FOMC) monetary policy meeting scheduled for October 27–28, he made clear that he would not preemptively project the Fed's future policy trajectory nor announce in advance what actions might be taken at the remaining meetings this year. He emphasized that each FOMC decision will be based on a case-by-case assessment of output, employment, and inflation, rather than following a pre‑set policy path, and reiterated his personal preference for the Fed to refrain from providing any forward guidance.
Notably, Walsh's series of remarks at the IMF reinforced the view that cutting-edge AI technologies will reshape the economy over the long term, yet he did not commit to pausing rate hikes in October. For global semiconductor stocks, which have recently faced sharp sell-offs and entered a correction, the positive signal is that he did not further bolster expectations of near-term rate hikes. However, whether U.S. Treasury yields can continue to ease, and whether AI‑related computing‑power stocks—already hit hard by recent selloffs—and the broader semiconductor and technology sectors can stage a rebound, will still hinge on whether inflation, corporate financing conditions, and the realization of AI‑driven earnings improve in tandem.
More noteworthy than his statement that he would "not preemptively judge monetary policy" is his assertion that, since the September FOMC meeting, economic developments have largely aligned with expectations, and that structural growth and a surge in investment are currently underway. This suggests that he has not yet equated the recent global tech‑stock correction—led by semiconductors—with a breakdown in the underlying logic of economic growth.
At the same time, he emphasized that most of the world is simultaneously tightening monetary policy, and that all countries must safeguard price stability. U.S. policy spillovers to overseas markets, while overseas policies, in turn, can feed back to affect the United States. The Fed's policy framework remains focused on anchoring inflation expectations and preserving institutional credibility, while leaving room for discretion. Prior to his remarks, MUFG's latest data showed that October rate‑hike probabilities in the interest‑rate futures market stood at around 27%, yet the market still fully priced in a 25‑basis‑point hike by year‑end. Consequently, both before and after Walsh's speech, trading in the rate‑futures market consistently reflected expectations of a possible delay in the timing of the next hike, rather than an end to the tightening cycle.
Wall Street giants have recently been converging on a forecast of "continued rate hikes and an extended tightening cycle," with the latest clear shift coming from Citi and UBS Group. The Federal Reserve raised rates by 25 basis points in September, and this round of institutional revisions is focused on how much further rate increases are needed. On October 14, Citi withdrew its prediction of "no further rate adjustments this year," instead projecting another 25-basis-point hike in December, followed by a pause until rate cuts begin in June 2027. Meanwhile, UBS Global Wealth Management, between October 12 and 15, extended its earlier outlook of "one more hike in December" to two 25-basis-point increases—one in December this year and another in March next year—bringing the total cumulative increase to 75 basis points, including the move already implemented in September. It also expects interest rates to remain unchanged through around the end of 2027.
At the heart of the reassessment of Wall Street's rate‑hike expectations is the fact that energy shocks are beginning to spread to a broader range of prices, while the economy has not yet weakened enough to compel the Federal Reserve to prioritize growth. In September, core CPI rose 0.3% month over month, exceeding market expectations of 0.2% for the second consecutive month, and the year‑over‑year pace accelerated from 2.4% to 2.6%. Rising costs in housing, airfares, automobiles, apparel, and technology‑related goods are making it increasingly difficult for policymakers to dismiss the inflation rebound as a one‑off oil‑price disturbance. Subsequently released data showed that the September PPI increased 1.3% month over month and 6.3% year over year, further heightening concerns about the pass‑through of upstream cost pressures to final‑demand prices.
AI First Drives Demand, Then Unleashes Productivity: Why Is the Computing Power Boom Coexisting with High Interest Rates?
In his remarks, Walsh characterized AI as the most profound economic transformation of his lifetime. Crucial to understanding this assessment is his simultaneous emphasis that AI's impact on the demand side will materialize before improvements on the supply side. From the perspective of data‑center engineering and investment in the AI‑compute‑chip industry chain, building compute clusters requires companies to first procure GPUs, TPUs, ASICs, HBM, and networking equipment, while also committing resources to power, cooling, facility construction, and financing. Meanwhile, the cost reductions and efficiency gains that arise once AI is widely integrated into R&D, manufacturing, and business processes necessitate deployment, organizational realignment, and commercial validation. This temporal lag enables AI to simultaneously boost semiconductor orders and increase capital and electricity demand, driving up certain prices and financing costs in the short term.
Accordingly, Walsh's AI‑driven optimism does not automatically translate into lower interest rates. While productivity gains can expand supply and curb unit costs, higher expected returns on investment may also intensify competition for capital, pushing real interest rates higher. As for the trajectory of 10‑year and 30‑year U.S. Treasury yields, markets must simultaneously factor in future policy rates, fiscal financing dynamics, and term‑premium compensation. Strong demand for computing power and persistently elevated long-term capital costs could well stem from the same wave of AI‑related investment expansion; accordingly, semiconductor earnings growth and valuation pressures can coexist.
His latest remarks also shed light on how policy research is tracking this shift: an in-person AI hearing is scheduled for November 16 in the San Francisco Bay Area, with a Federal Reserve working group developing new insights to inform FOMC deliberations. AI is expected to enhance central banks' ability to interpret data, though its short-term implications for employment remain uncertain. As for private‑sector technology investment, the Fed will refrain from arbitrating right or wrong and instead focus on financial tail risks that can be addressed through policy. These signals suggest that the Fed and other global central banks are placing greater emphasis on AI's contribution to growth, while also assessing financing structures and risk transmission—without committing to safeguarding AI‑related asset prices.
U.S. Treasury yields have already eased, but the semiconductor sector's rebound will still require a dual boost: lower credit‑financing costs and improving earnings.
Long-term yields have indeed begun to ease. According to the U.S. Treasury's standardized measures, on October 15, the yields on 10-year and 30-year Treasury bonds stood at 5.10% and 5.43%, respectively—both down 9 basis points on the day. Year-to-date through the 15th, they have fallen by 14 and 17 basis points, respectively, compared with declines of 19 and 21 basis points from late September. Nevertheless, both remain 35 and 18 basis points higher than at the end of August, indicating that the recent pullback has only partially reversed the earlier surge in borrowing costs.
Notably, a key driver of this recent pullback has been risk‑aversion buying of U.S. Treasuries, triggered by volatility in AI‑related stocks and the credit markets. According to TD Securities, investors are revising down their growth forecasts, causing U.S. Treasuries to once again serve as a hedge against equities. This has created a seemingly paradoxical market dynamic: pressure on computing‑power stocks is pushing Treasury yields lower, yet that yield decline has so far failed to allay concerns about financing for compute‑intensive projects. Corporate borrowing costs comprise both benchmark interest rates and credit spreads; if the former falls while the latter widens, data‑center projects reliant on external financing may still face persistently high effective borrowing costs.
Global stock markets were also in a post‑crash recovery phase as Chairman Powell delivered his remarks. As of October 15, the Philadelphia Semiconductor Index had fallen 0.88%, marking a cumulative decline of roughly 12.4% over seven consecutive sessions; this week's drop stood at about 7.9%, and the month-to-date decline was around 8.3%. Despite still being up roughly 63.5% year to date, short‑term momentum has clearly weakened. On October 16, during the Asian midday session (Beijing time), South Korea's benchmark KOSPI index—often dubbed the "barometer" for AI‑related memory chips—rebounded sharply by 3.17% in a single day, with Samsung and SK Hynix both rallying in tandem. This rally preceded Powell's speech, but its upward trajectory has since persisted.
Wash's remarks this time have, to some extent, provided support for restoring sentiment in the battered global semiconductor sector. However, the true catalyst for a sustained rebound in semiconductors lies in the confluence of "a decline in long-term yields, stable credit spreads, and continued fulfillment of computing‑power orders and cash‑flow commitments." If financing pressures ease while AI demand remains robust, the market will be poised to once again assign higher valuations to growth; among them, leading chip and equipment companies with strong cash flows, stable customer funding sources, and robust order‑conversion capabilities are likely to benefit first. Wash's emphasis on the long-term transformative impact of AI offers a macroeconomic rationale for this growth narrative, but whether the near‑term rally can evolve into a sustained trend will depend on whether lower funding costs can genuinely translate into increased industrial investment.
Wall Street financial giant Goldman Sachs recently described "three major reinforcements" — share buybacks, retail investors, and foreign capital — which are expected to improve the funding environment for AI assets to regain pricing. In its latest report, "Witching Hour," Goldman Sachs' capital flows team notes that, according to their calculations, by late October roughly 48% of S&P 500 constituents will once again enter the buyback window; historically, November accounts for about 10.1% of the year's total buyback activity. Coupled with the waning year-end tax‑selling pressure from mutual funds, year‑end retail inflows, and post‑midterm election overseas allocation demand, the supply‑demand balance for equities could gradually improve. Notably, approximately $190 billion in overseas buying represents a potential scale extrapolated from historical trends. For semiconductors, which have experienced concentrated deleveraging, a resumption of buying could trigger position rebuilding even if companies merely demonstrate that orders and earnings have not deteriorated significantly. However, buyback execution in the tech sector has declined 40% year over year, indicating that capital expenditures are competing with shareholder returns for available funds. For AI stocks to continue attracting incremental allocations, they must still prove that capital investments in key AI infrastructure—such as GPUs, network‑infrastructure and data‑center optical interconnects, and CPUs—can generate higher future cash flows.
Editor/lambor