Is the AI frenzy peaking? Michael Hartnett, a strategist at Bank of America, made a bold prediction in his latest report: the "best trade" heading into 2026 will be shorting the corporate bonds of "hyperscale cloud providers" heavily invested in AI. He argues that AI-driven debt burdens will become the new "Achilles' heel" for these tech giants.

Hartnett’s view hinges on a key observation: the easy financial conditions fueling the AI boom are nearing an inflection point. While central banks globally cut rates 167 times over the past 12 months, the number of expected rate cuts for the next year has dropped sharply to 81. This waning liquidity momentum suggests credit spreads have bottomed—bad news for capital-intensive sectors.

Despite cracks emerging in credit markets, Hartnett believes the stock market rally won’t end abruptly. He expects asset allocators to remain broadly long equities into 2026, supported by factors like the "Fed put" and the "Trump put."

Hartnett identifies a critical date: May 15 next year, when the new Fed chair is appointed. He argues that before then, significant "risk-off" signals triggered by bank stocks or credit spreads are unlikely. In other words, staying long equities seems safe until May.

**AI Debt Risk: Why Short Cloud Giants?** Hartnett’s bearish thesis centers on shifting financial conditions (FCI). He notes that the "secret sauce" for risk assets in 2025 was "massive easing," with 167 rate cuts globally. But this momentum is fading, with only 81 cuts expected in the next 12 months.

He believes the bulk of the easing cycle is over. As liquidity tightens, concerns about credit cracks and financing for AI-driven capex surges are growing.

The immediate impact is pressure on credit markets. Hartnett observes that peaks in financial easing typically coincide with troughs in credit spreads. Today, tech firms’ massive AI-related capex—exceeding cash flow support—has widened bond spreads and credit default swaps (CDS). He calls this "trouble in parade."

For example, Oracle’s CDS recently surged above 100bps after reports highlighted its weak credit profile in early October.

Hartnett even quipped: "When the Fed next launches QE, you’ll know they’re buying AI cloud giant bonds."

**Unaffordable Borrowing Costs** Hartnett starkly highlights a core contradiction in the U.S. economy: while "massive easing" fueled Wall Street’s boom, Main Street (the U.S. government) faces "unaffordable" borrowing costs.

He contrasts key rates: "The U.S. government borrows at 4%, investment-grade (IG) corporates at 5%, mortgages exceed 6%, small-business credit lines cost 7%, home equity loans 8%, auto loans 13%, SBA loans 14%, and credit card APRs hit 20%." This disparity underscores uneven monetary transmission and the reality of a "K-shaped" recovery.

**Goldilocks Narrative: Risk-On Until May** Despite credit worries, Hartnett argues the macro narrative supporting markets remains strong.

He frames it as a "Goldilocks" scenario: "lower rates/higher profits, the U.S. government pushing for pre-midterm growth, equities mirroring the economy with K-shaped consumption, and AI curbing inflation via job cuts and productivity gains."

Thus, he expects "2026 price action to be front-run." Asset allocators will stay long stocks until bank stocks or credit spreads flash clear risk signals—unlikely before May.

This timing isn’t arbitrary. May 15, 2026, marks the new Fed chair’s swearing-in, with markets anticipating a more dovish leader. Until then, expectations of the "Fed put," "Trump put," and "Gen Z put" will sustain equity demand—pricing in future policy easing while overlooking credit risks.

**Beyond AI: Watch PMI Acceleration and Cooling Inflation** Hartnett also spots macro opportunities. Tax cuts, rate reductions, and U.S. reshoring could drive PMI toward 55 (expansion territory).

Under this "PMI trade," he favors commodities, globally PMI-sensitive markets (China, Japan, Germany), and U.S. small/mid-caps. He notes "Doctor Copper" (a growth barometer) aligns with this view.

U.S. small/mid-caps, trailing in 2025 with a 14x P/E (vs. S&P 500’s 23x), look ripe for catch-up.

But Hartnett warns: weak rebounds in early-cycle sectors (real estate, construction, retail) may signal AI’s negative job impact, potentially capping small-cap and industrial gains.

He also proposes a contrarian "inflation trade": if U.S. core CPI falls to 2%, long-dated Treasuries (e.g., ZROZ) could rally. Politically, he ties this to midterms: "Whoever wins the 'affordability' issue wins the election."

To woo voters, the U.S. may directly intervene in energy, healthcare, and housing costs—hurting sector margins. In this scenario, long-Treasury trades have upside.

Yet, Hartnett sets a clear exit: "Take profits before May 15" to avoid policy uncertainty under the new Fed chair.