The fate of most of the US stock market rests on a single question.
Will the huge amounts being poured into artificial intelligence eventually deliver solid returns for users, or will the absence of payoff cause the investment surge to collapse?
We don’t have a definitive answer. However, for bonds the outcome appears irrelevant.
We believe either scenario would still push interest rates lower. The unwinding of AI’s upward pressure on rates is a key reason we anticipate rates to decline in the coming years, which would create strong returns for inflation.
Bond yields surged to multidecade highs a month ago. They have since retreated from those peaks but remain markedly high, with the 10‑year Treasury yield up almost one percentage point year‑to‑date.
We forecast that the Federal Reserve will resume cutting rates in 2027, eventually lowering the federal‑funds target to 2.5%‑2.75% by the end of 2028—about 1.25 percentage points below current levels. This should pull the 10‑year Treasury yield toward an average of 3.5% in 2029 and beyond, down from today’s 5.1%.
The first route is that AI begins to generate a clear return on investment. By “investment” we mean the spending by AI end users. For instance, a consumer‑goods firm might purchase an AI model like ChatGPT to automate routine accounting.
A “return” simply means the user creates enough additional profit to more than cover the AI expense. The core of the issue is that this profit will largely stem from labor‑cost savings.
Untangling the myriad opinions on this topic would require a separate article. Some argue that the biggest AI adopters will boost revenue rather than cut jobs. Yet, even if labor‑light AI firms outpace labor‑heavy laggards, the net effect would still be widespread layoffs.
How extensive could those layoffs be? To cite a representative estimate, Goldman Sachs
