AI Trading Cools Down, Safe-Haven Buying Flows into US Treasuries! "The Anchor of Global Asset Pricing" Faces Its First Monthly Decline in 8 Months

According to Zhihu Finance APP, during Friday’s Asian trading session, the rally in U.S. Treasury bonds accelerated further. Major sell-offs occurred in technology stocks focused on themes such as AI computing power and semiconductors within the U.S. and global markets, prompting investors worldwide to turn to U.S. government bonds for a safe haven. Yields on all U.S. Treasuries fell by at least 2 basis points, extending the downward trend from the New York trading session. Australian, Japanese, and New Zealand bonds followed suit, with each country’s benchmark 10-year government bond yields declining by about 5 basis points.

Disappointing earnings and outlooks from tech giants focused on the AI cloud computing sector, including Alphabet (Google’s parent company) and Amazon—especially cash flow data that fell well short of market expectations—significantly intensified market skepticism toward AI trading themes and reinforced pessimism about a potential AI bubble burst, accelerating the flow of funds into safe-haven assets. Meanwhile, Binance, the world’s largest cryptocurrency exchange, was reportedly targeted by a hacker attack, further heightening risk-off sentiment. The benchmark indicator for U.S. Treasury yields (specifically, the 10-year Treasury note, often referred to as the “anchor for global asset pricing”) has dropped nearly 40 basis points in October, likely to record the first significant monthly decline since February.

Kenneth Crompton, Head of Rates Strategy at National Australia Bank in Sydney, said: "After this week's tech company earnings releases, a fresh wave of risk aversion is again driving up bond prices and pushing down yields this morning."

During the Friday Asian trading session, Nasdaq futures declined over 2%, with Amazon and Intel sliding 7.7% and 23% respectively in after-hours trading. Nvidia dropped 2.5%, Micron Technology fell 4.4%, and Oracle slipped 4%. Intel, in particular, plunged as much as 23% in after-hours trading due to its disastrous guidance. At the time of writing, South Korea’s KOSPI index—heavily weighted towards memory chip giants SK Hynix and Samsung (collectively over 50%)—was down more than 5.6%, prompting an intraday “sidecar” mechanism halt on the KOSPI by the Korea Exchange; Japan’s Nikkei 225 fell nearly 3%.

As shown in the chart below, U.S. Treasuries are on track to post their best monthly return since February 2025. The Bloomberg U.S. Treasury Index, which tracks returns in the Treasury market, has climbed 2% in October, poised for its biggest monthly gain since February 2025.

Throughout the past month, deteriorating sentiment toward AI-related stocks has been a major driver of gains in U.S. government bonds. Federal Reserve Chair Kevin Walsh stated that future rate decisions would depend on subsequent economic data rather than any preset hawkish path. This stance has provided strong support for the bullish short-term Treasury price outlook.

Traders are closely monitoring policy prospects. Despite high energy costs increasing price pressure, both the Federal Reserve and the Reserve Bank of New Zealand held interest rates steady this week, and traders unanimously expect the Bank of Japan to follow the Fed’s lead and stand pat as well.

As the U.S. midterm elections approach, the bond market faces another looming risk. Investors are increasingly skeptical whether a potential political stalemate in Congress—a familiar scenario in recent decades—can again ease their concerns over the trajectory of U.S. fiscal policy.

The day after the election, the U.S. Treasury will announce its quarterly refinancing plan, which the market views as laying the groundwork for any major adjustments to debt management policy.

However, the present market focus remains on the sell-off in tech stocks: Investors are increasingly doubting whether massive AI spending can generate sufficient profit growth to support current valuations.

Andrew Ticehurst, Senior Rates Strategist at Nomura Holdings in Sydney, commented: “Ongoing concerns about the valuations of hyperscale cloud providers—combined with deeply disappointing earnings reports from Alphabet, Amazon, and Intel—are driving a portion of risk-off capital into the bond market.”

He added that the bond rally may continue, as “we don’t think Walsh sounded especially hawkish in the post-meeting press conference, and these AI-related concerns seem likely to persist.”

AI Trading Under Pressure, U.S. Treasury Market Quickly Steps In: Global Capital Revalues Cash Flow Certainty

As AI computing power and semiconductor stocks come under selling pressure, U.S. Treasuries have once again demonstrated their value as a hedging asset in investment portfolios. Technology investors are re-evaluating the return cycle of massive capital expenditures, with some funds shifting toward government bonds featuring higher liquidity and more predictable cash flows. During Asian trading on October 30, the Treasury rally extended, with the 10-year yield dropping to around 4.90%; Australian, Japanese, and New Zealand sovereign bonds simultaneously rallied, reflecting that this wave of risk-off demand has extended beyond the U.S. market.

Bloomberg bond market data shows that in early Asian trading, yields on all U.S. Treasury maturities declined by at least 2 basis points, while 10-year government bond yields for Australia, Japan, and New Zealand each fell by about 5 basis points. The U.S. benchmark 10-year Treasury yield has dropped close to 40 basis points in October, setting up potentially the first monthly decline since February this year. Meanwhile, the Bloomberg U.S. Treasury return index has climbed roughly 2%, also aiming for its best monthly performance since February 2025. The former measures changes in yields, while the latter tracks bond investment returns. Both sets of data point to one outcome: U.S. Treasuries are absorbing robust risk-off demand triggered by declining risk appetite for technology stocks.

The key to this round of asset rotation is the market’s shift from “AI demand growth” to “actual cash returns available to investors.” From an industrial perspective, growing cloud revenues require a process of server deliveries, datacenter power provisioning, network interconnections, increased utilization, and client payments before they can be converted into cash flows covering construction and financing costs. Robust orders for GPUs, HBM, and liquid cooling equipment prove construction demand exists but cannot alone demonstrate that each project’s capital return is sufficient to cover funding costs. As investors demand higher conversion of cash flows, both stock valuations and corporate credit risk premiums may come under simultaneous pressure, while government bonds receive support from flight-to-safety flows.

This also explains why technology stocks can fall as Treasuries rise: What the market is discounting is trust in companies’ future profitability and financing capability, while some funds are increasing allocations to government bonds. In this adjustment, U.S. Treasuries have regained their role in hedging equity drawdowns. However, this protective ability still depends on the shock’s source—a concern over earnings and credit typically benefits safe-haven buying, while out-of-control energy inflation could depress both equity and long-duration bond prices at the same time.

U.S. Treasury Secretary Baycent Reassures on Financing, Market Remains Selective on Risk: Can Treasury Gains Help Lower AI Costs?

Baycent’s latest views are not necessarily at odds with current market moves. He emphasized that new cloud companies are only part of the AI ecosystem, that financing transactions are still taking place, and cited Meta’s use of equity financing as an example showing capital markets can still provide construction funds. This answers the question “Are financing channels still functioning?” Meanwhile, investors selling tech stocks and buying Treasuries are evaluating “Whether current prices sufficiently compensate for earnings, cash flow, and credit risk.” Successful financing does not automatically mean cheap funding, nor does it guarantee that every project will generate adequate returns.

From a financing mechanism angle, pricing of newly issued corporate bonds typically includes the corresponding maturity’s benchmark yield, plus compensation for credit, liquidity, and other risks. Thus, a decline in Treasury yields can lower the benchmark part of funding costs, but if the credit spreads for firms like CoreWeave widen more rapidly, their overall funding costs might still rise. For cash-rich cloud giants, the pressure mainly shows up in required returns on new investments and discounted equity valuations; for new cloud companies reliant on ongoing external financing, the stress also directly impacts construction progress, equipment procurement, and refinancing capacity. Therefore, stronger Treasuries may primarily reflect a flight to quality, rather than a broad-based improvement in the AI funding environment.

The Fed’s pause has provided room for Treasury buying, yet has not declared an end to tightening. On October 28, the Fed voted 10-2 to maintain the target range for the federal funds rate at 3.75%–4%, with two dissenters calling for a 25 basis point hike. Walsh acknowledged that parts of business and household finance and credit conditions have tightened, while stressing that the overall financial environment cannot yet be described as tight, and that further policy action requires more information. This supports the view that no extra hawkish “market jitters” signal was released, but does not justify expectations of imminent rate cuts.

Baycent’s comments on energy align with another potentially longer-lasting favorable path: If geopolitical tensions ease and oil supplies recover sufficiently, inflation and borrowing costs may decline. He also said future rate hikes may not constitute a policy error. Taken together, his statements indicate that the Treasury’s optimism is primarily based on economic resilience and the alleviation of energy pressures, not on immediate monetary easing to prop up asset prices.

For portfolio allocation, Treasury duration and corporate credit require separate assessment: Holding Treasuries allows participation in gains driven by falling yields, while buying AI company bonds also requires bearing the risk of credit spreads expanding further. In the next phase, longer-dated Treasuries will still face post-midterm fiscal expectations and the issuance plans revealed in the November 4 quarterly refinancing announcement. Risk-off demand can support current bond prices, but long-term supply, record-setting fiscal deficits, and inflation risk will continue to affect the term premium demanded by investors.