Evercore strategists outlined specific credit and bond market signals that may precede a decline in artificial intelligence equities after the recent tech sell-off. Julian Emanuel's team published a research note on Sunday describing the recent price drops as the start of a valuation reassessment for AI-related investments.
The firm identified four key metrics to watch. First, investment-grade credit default swap (CDS) spreads must exceed their March peak of 68.33. Second, high-yield CDS spreads need to rise above their previous high of 411.04. CDS contracts protect against borrower defaults, so widening spreads indicate investors demand more compensation for credit risk. Third, analysts' earnings-per-share revisions must reverse their upward trend, signaling lower profitability expectations. Fourth, the yield on 10-year US Treasury securities must surpass 5.62%.
Evercore's Washington policy team labeled the 5.62% yield threshold a "fiscal alarm bell." Citing a recent peak of 5.36%, the team called 5.62% a level where market attention to US fiscal conditions would likely intensify. Higher yields increase borrowing costs and affect the discount rates used to value future corporate earnings, a factor particularly relevant for tech firms relying on long-term growth projections.
The strategists challenged the assumption that hyperscalers can issue debt without market constraints. They referenced an estimated $1.3 trillion in AI-related investment by 2027, highlighting the financing needed for data centers and computing equipment. Evercore dismissed the idea that companies could sustain borrowing programs regardless of market conditions or negative free cash flow as "pure fantasy."
To contextualize these risks, the team compared current expectations to past financial instability events, including the 2000 tech bubble collapse, the 2018 equity decline, and Japan's 1989 asset bubble. These comparisons assessed financing risks rather than predicting a repeat of those events. The note stated that last week's decline reinforced the "this time is no different" mindset amid a "perfect storm" for the AI bull market.
Despite these warnings, Evercore maintained its year-end 2026 S&P 500 target of 7,750. The firm advised investors to keep exposure to its "AI class of 2026" stocks while indicators remain below warning levels. Strategists recommended maintaining hedges for the November 3 US midterm elections using negative-beta equities, S&P 500 straddles, and December Nasdaq-100 put spread collars.
The report also suggested technical levels for adjusting positions. Investors could increase equity exposure and remove hedges if the Nasdaq-100 tests its 200-day moving average at approximately 674 or if the S&P 500 reaches 725. These figures appear inconsistent with prevailing index levels and may refer to another measure or require clarification.