An overwhelming majority of corporate discussions about productivity that mention AI are projections about potential benefits, not evidence of current performance gains. According to a recent analysis by the St. Louis Fed, 95% of productivity-related sentences that include artificial intelligence in earnings calls are framed around future outcomes. This suggests that businesses are still largely in the early stages of incorporating AI, with most of their focus on what might be achieved rather than what has already been realized in concrete terms.
AI investment is accelerating
The pace of investment in artificial intelligence is rapidly increasing, driven primarily by demand for the hardware and services needed to power the technology. Goldman Sachs has predicted that U.S. companies will invest nearly $600 billion in AI next year, which equates to about 2% of the nation’s gross domestic product. This spending is concentrated in infrastructure such as advanced semiconductor chips, data centers, and cloud computing platforms. While these developments are crucial and generate strong profits for tech suppliers, they represent just the first phase of a broader transformation. The real impact of AI is expected to emerge only when companies start deploying the technology in meaningful ways to reduce costs, enhance efficiency, and restructure entire sectors of the economy.
Talk is cheap
Language used in corporate communications has changed significantly as AI has become more prominent. In the early 2000s, businesses hardly ever referenced artificial intelligence when discussing productivity. More recently, however, terms like “using AI” or “AI tools” have begun to replace the once-popular phrase “generative AI.” This evolution in terminology reflects a transition from vague speculation to a more practical orientation. The decline of “generative AI” as a buzzword signals a growing emphasis on tangible applications over abstract concepts. Nevertheless, the change in language is not yet matched by widespread operational adoption.
The St. Louis Fed economists analyzed more than 490,000 transcripts from corporate earnings calls covering 5,198 publicly traded U.S. companies between 2000 and 2025. Their findings indicate that the great majority of discussions about AI and productivity remain speculative. These discussions are typically cast in future tense, signaling anticipation rather than realization. This trend mirrors similar patterns from earlier technological revolutions, where initial excitement and investment preceded measurable improvements in productivity. While AI is currently in the experimental phase, businesses are making significant investments and preparing for future outcomes. However, the technology has yet to deliver the kind of consistent, long-term gains that would mark the transition from expectation to actual performance. Until those results become evident, AI’s promise remains largely in the realm of potential.
