AI frenzy faces the “funding cutoff line” test? Renowned Wall Street strategist warns: With the US dollar and Treasury yields not peaking, risk appetite remains difficult to recover

According to Odaily News, the team led by Michael Hartnett, a senior strategist at Bank of America and known as "Wall Street's Most Accurate Strategist," released a research report stating that before the dollar rally shows significant signs of peaking and before the 10-year and longer-term U.S. Treasury yields—pushed up by energy inflation—retreat from historic highs, risk assets are unlikely to escape the forces of deleveraging and sell-off pressure. It is noted that the Bloomberg Dollar Index has rebounded about 3% from its September low, reflecting investors selling off equities, cryptocurrencies, and other risk assets while accumulating cash; at the same time, the “anchor of global asset pricing”—the U.S. 10-year Treasury yield—hit 5.34% on October 1, marking a 24-year high (since 2002), before pulling back.

Therefore, while remaining cautious, Hartnett suggests investors start buying bonds that have been neglected by the market and puts forward a policy support expectation: if yields continue to rise and threaten the AI investment boom ahead of the mid-November elections, the U.S. government may increase its Treasury buyback efforts. He is particularly focused on whether the decline in bank stocks will fully spread to small and mid-cap stocks in the equity market, which would indicate a severe shake in economists’ and investors’ optimism about a “soft landing” for the U.S. economy—with the resulting huge sell-off pressure potentially spreading to tech stocks.

According to Hartnett’s team of strategists, investment sentiment along the AI computing power industry chain still faces a dual stress test from the dollar and the long-end yield curve. Whether the explosive growth in AI computing demand can translate into rising asset prices also depends on whether financing conditions and valuation discount pressures can abate simultaneously.

The recent surge in 10-year and longer U.S. Treasury yields driven by Middle East energy inflation represents a situation of "diplomatic and military pressures in tandem, oil prices correcting but geopolitical war premiums remaining." Qatar continues to broker the U.S.-Iran “seven-day confidence-building plan,” with differences in the sequencing of actions; the U.S. is moving more military force to the Middle East, while Iran is also preparing for potentially renewed large-scale U.S. attacks.

As of 16:40 Beijing time on October 2, the international crude oil pricing benchmark—Brent crude futures—was last quoted at $99.48 per barrel, down 2.77% on the day, while WTI crude futures were at $89.52 per barrel, down 3.61%. Calculating from the settlement prices of $72.48 and $67.02 on February 27—the last trading day before the outbreak of war on February 28—these two benchmarks are still up about 37.3% and 33.6% respectively. This comparison, using the nearest month futures prices at various points in time, shows that even after the recent pullback, energy prices are still significantly higher than pre-war levels.

Bank of America Senior Strategist Michael Hartnett: Risk Aversion May Persist Until the Dollar Index Peaks

Michael Hartnett, senior strategist at Bank of America, said his team believes that until the recent sharp rise in the dollar shows signs of peaking, investors will continue to avoid higher-risk trades.

In addition to waiting for a key signal that the dollar has peaked, Hartnett’s team stated in a recent report that the market's unease and anxiety-driven selling pressure may persist until rising bond yields, at over 20-year highs, retreat. He advises, “Buy the assets most scorned by the market—the ones that have seen major recent sell-offs,” and is beginning to favor adding some long-dated U.S. Treasury assets to portfolios that have suffered continued selling recently.

As investors in financial markets exit higher-risk asset positions and start rebuilding cash buffers, the Bloomberg Dollar Index is up 3% from its September low. Meanwhile, bond yields have also risen, driven by factors such as inflation pressures from the Iran conflict, expectations for further monetary tightening, and strong corporate profit growth.

As shown in the chart above, the dollar index and Treasury yields have surged recently while the stock market rally has stalled.

Hartnett noted that although recent price action shows the market is lowering leverage and risk exposure to equities, cryptocurrencies, and other assets, larger-scale Treasury buybacks by the U.S. government could provide downside support for the market—especially if rising yields threaten the U.S. AI investment boom ahead of the November midterm elections.

Hartnett said that if small and mid-cap stocks also join banks in sharp declines, the downside risk will become even more worrying. He emphasized that this would signal the market's optimism about strong economic growth has peaked and could eventually drag down tech stocks.

From Energy Transport Bottlenecks to AI Craze’s "Financing Kill Line"

A series of cautious views recently put forward by Bank of America senior strategist Michael Hartnett have always revolved around capital, positions, and the bond market. On September 11, his team pointed out that U.S. equity funds had a net outflow of $14.2 billion in the previous three weeks, while weekly inflows into global equity funds fell from $52 billion in July to $7 billion.

The subsequent Bank of America September fund manager survey showed cash allocations rose to 3.9% but remained at levels Hartnett sees as triggering contrarian sell signals in risk assets, and that the disorderly rise in bond yields became the tail risk most feared by respondents. On September 25, he also warned that the U.S. Treasury volatility index MOVE surged 33% in two days. The core of these latest views means: even if earnings and economic growth driven by the AI theme remain resilient, thin cash buffers, rising bond volatility, repeated highs in Treasury yields, and a strong dollar could constrain investors’ risk appetite.

Energy transportation is being repaired, but there is still some way to go before normal shipping costs return. Saudi Arabia’s east-west pipelines are reinstated, Yanbu terminal has resumed loading, increasing export routes that bypass the Strait of Hormuz; although the design capacity is 7 million barrels per day, Reuters data on September 29 indicated that actual deliveries remain at around 2–2.65 million barrels per day.

While some statistics show an increase in Hormuz traffic with 19–21 batches of LNG shipped out in September, three oil tankers were still attacked by unidentified ordnance on September 29; security for Bab-el-Mandeb is also required, with the French military reporting on October 1 that about 10 merchant ships were escorted in the past week. Energy route differences among Gulf oil producers are particularly critical: crude from Yanbu north to Europe can use the Suez Canal, but shipping south to Asia still generally passes Bab-el-Mandeb, so insurance, escort, and rerouting costs continue to hamper transportation efficiency. The volume of supply recovery and how cheaply energy can be delivered are two variables the market must price in at the same time.

The "financing kill line" of the AI craze is drawing closer. Against a backdrop of energy inflation driving the 10-year U.S. Treasury yield to a more-than-20-year high and a surging dollar sending risk aversion upward and dampening appetite, the "financing kill line" for the AI investment frenzy seems ever nearer—that is, as the "anchor of global asset pricing" 10-year Treasury yield hits a new high since 2002 and benchmark funding costs remain near historic peaks, financing progress critical to AI capital spending and infrastructure project returns is beginning to be constrained. The market is increasingly questioning whether many large AI data center projects can continue to deliver expected returns with rising capital costs.

When the expected cash returns of new computing power projects—after deducting costs such as electricity, operations, and equipment upgrades—can no longer cover overall cost metrics including financing, further expansion ceases to create economic value and may mark the start of failure for individual large-scale AI infrastructure projects; it will impact those with weaker cash flow and unsecured financing first.

The 10-year U.S. Treasury is called the "anchor of global asset pricing" because of its benchmark role in dollar financing systems and mid- to long-term cash flow valuations. The U.S. Treasury market is vast and actively traded, with the dollar widely used for international financing and reserves, so changes in its yield have cross-market effects—the yield on dollar corporate bonds is typically based on similar-tenor Treasuries plus a credit spread, mortgage rates are influenced by Treasuries and MBS pricing, and stock and real estate valuations are highly sensitive to future cash flow discount rates. When this benchmark rises while earnings and rent expectations do not improve, asset prices come under downward pressure. The effects are also transmitted overseas via the dollar’s funding cost, FX hedges, and cross-border capital flows; different currencies, tenors, and credit risks determine the specific asset impact.

From a data center project perspective, GPU servers, power access, and cooling infrastructure require upfront investment, but computing power service income is collected over time; rising long-term risk-free rates and credit spreads drive up funding costs and lower forward cash flow valuations, and a stronger dollar increases the repayment and procurement burdens for non-dollar borrowers. Therefore, robust computing power demand and tighter project financing can occur simultaneously, with the first to be tested being expansion plans that rely on external financing and have distant cash returns. The policy support Hartnett expects is intended to ease these capital cost constraints: the Treasury Department has expanded liquidity support repo operations for long-term Treasuries, but the official stated purpose is to improve market liquidity; further ramping up to protect AI investment remains Hartnett’s policy judgment. For investors’ overall strategies, the performance of the dollar, the yield on 10-year and longer U.S. Treasuries, and the market performance of banks and small/mid-cap stocks are increasingly becoming important signals to test whether the AI computing power bull market can spread to a broader range of stock sectors.