Bank of America Securities Chief Investment Strategist Michael Hartnett has issued a stark warning that surging global bond yields, now sitting at two-decade highs, represent the most significant threat to the ongoing artificial intelligence capital expenditure boom. He further cautioned that the upcoming US midterm elections could serve as the catalyst that ignites a broad market sell-off.

In his latest Flow Show weekly report, Hartnett projected that a Democratic sweep of both chambers of Congress in the midterms would likely trigger an equity market decline exceeding 10%, accompanied by a softer US dollar and lower bond yields. Such an outcome, he argued, would also expose the fragility of the current AI investment bubble, which he identified as one of the most significant tail risks currently underpriced by investors.

According to Polymarket data, the probability of a Democratic sweep has climbed to 50%, significantly outpacing the 10% likelihood assigned to a Republican sweep. President Trump's approval rating, currently hovering between 35% and 40%, sits markedly below the historical average of 53% for presidents at this stage before midterm elections. This political backdrop is reinforcing Hartnett's warnings about potential market volatility.

Bond Market Raises the Alarm First

Hartnett suggested that the most significant market development last week was not the stronger-than-expected jobs report, but rather the comprehensive breakdown across global bond markets. The yield on the 10-year US Treasury climbed to 4.81%, approaching levels last seen during the 2008 financial crisis, while the 30-year yield reached 5.31%, its highest point since 2007. Meanwhile, Japan's 10-year government bond yield broke above 3.0% for the first time since 1996, with the 30-year yield hitting 4.2%—roughly four times the Bank of Japan's policy rate. In Europe, Germany's 10-year yield rose to 3.38%, a post-2011 high, while spreads between French and Italian bonds versus German bunds widened to 88 and 84 basis points respectively, reaching levels associated with the 2012 eurozone debt crisis. The Bloomberg Global Bond Yield Index has climbed to its highest point since 2007, sitting just one percentage point below its record peak for this century.

Hartnett distilled this phenomenon into a core thesis: "a bond-dominated bubble." He argued that longer-dated yields, rather than equity narratives, are the true anchor for the AI trade. He noted that until global 30-year yields fall below 5%, the builders and financiers of AI infrastructure will continue to underperform the companies that are applying AI technology.

The Midterm Elections as an Underappreciated Market Variable

Hartnett clarified that these midterms do not represent a "regime change" election comparable to Thatcher/Reagan in 1980 or the Brexit/Trump victories of 2016, and they will not fundamentally alter the upward trajectory of US government spending. However, he emphasized that the structural differences in possible outcomes could have a notable impact on asset prices. According to the August BofA Fund Manager Survey, 47% of respondents anticipate a divided government with Republicans controlling the Senate and Democrats controlling the House, 23% expect a Democratic sweep, and only 9% foresee Republicans retaining control of both chambers. Currently, Republicans hold a 53-47 edge in the Senate and a 218-212 advantage in the House.

For Democrats to achieve a sweep, they would need to capture at least four of six vulnerable Republican-held Senate seats, including North Carolina (92% flip probability), Maine (69%), Alaska (64%), Ohio (55%), Texas (51%), and Iowa (37%), while also defending their own vulnerable seats in Georgia (94% hold probability), New Hampshire (84%), and Michigan (65%). Hartnett specifically highlighted Ohio, Texas, Iowa, and Michigan as key battleground states that investors should closely monitor.

Wall Street is also paying close attention to the Texas gubernatorial race, where Republican incumbent Governor Abbott (49% polling support) faces Democratic challenger Hinojosa (45%). This contest is being viewed as a significant indicator for the future of AI data center expansion policy, particularly after Abbott recently announced a moratorium on new data center construction in response to declining poll numbers.

Market Impact of a Democratic Sweep

Hartnett outlined a clear transmission mechanism for how a Democratic sweep would impact markets. He argued that a shift from "populist capitalism" to "populist socialism" would reverse the direction of tax and regulatory policies, creating a negative shock to corporate earnings. This would be accompanied by policy initiatives aimed at lowering inflation, improving healthcare access, and mitigating K-shaped wealth inequality, which would directly challenge the AI capital expenditure boom and the "too-big-to-fail" Wall Street ecosystem. Furthermore, the erosion of Trump's political capital would weaken his ability to execute on priority issues such as AI, resource monopolies, and diplomatic pressure tactics.

Under this scenario, Hartnett recommends shorting financial stocks and the US dollar as the optimal hedges. He projects an equity market decline exceeding 10%, a weaker dollar, and lower bond yields. International equities would likely outperform, with Europe favored over Asia.

Conversely, if Republicans unexpectedly retain control of both chambers, it would signal a return to broad risk appetite, giving the AI bubble a green light and reigniting the narrative of US dollar exceptionalism. The most probable outcome—Republicans controlling the Senate and Democrats controlling the House—would correspond to moderate risk appetite, with gridlock being viewed as "goldilocks" conditions.

Strategic Allocation: Favor Commodities and Gold, Beware Crowded AI Trades

Within this macro framework, Hartnett maintains his core cross-cycle recommendation: go long commodities and gold as hedges against inflation and geopolitical risks. He pointed out that governments' "whatever it takes" fiscal interventions are suppressing long-end yields and supporting nominal GDP growth, which reinforces the strategic logic that "any asset is better than bonds."

Regarding the AI trade, Hartnett was more direct in his caution. He noted that the free cash flow of hyperscale cloud providers has turned negative under the pressure of capital expenditure commitments, making the AI bubble "prone to bursting at any time." He proposed a post-bubble operational framework of "long humiliation, short arrogance," suggesting a shift toward longer-duration bonds and defensive sectors—including consumer staples, mining/materials, and healthcare—while avoiding crowded trades in AI infrastructure.

From a longer-term cyclical perspective, Hartnett also highlighted a contrarian signal: the US stock market has delivered a 15% rolling 10-year return, commodities 11%, and US bonds negative 2%—the worst performance in nearly a century. Historical data shows that when long-term bond returns turn negative, it often presents good buying opportunities for stocks (as in 1939, 1974, and 2009) and commodities (1933 and 2018), providing a historical basis for his tactical bullish stance on Q4 bonds.