U.S. AI adoption slips for the first time as markets soar, raising talk of an AI winter
For the first time since late 2023, U.S. businesses are pulling back on artificial intelligence. Fresh data from the U.S. Census Bureau’s biweekly Business Trends and Outlook Survey (BTOS), which tracks 1.2 million firms, shows AI usage dipped in August—led by large enterprises—even as tech stocks continue to press record highs.
The reversal is most visible among companies with more than 250 employees: AI adoption fell from 13.5 percent in June to 12 percent in August, breaking a steady climb that began in November 2023. While overall use of AI to produce goods and services is still higher than it was late last year—rising from 3.9 percent in late 2023 to more than 5 percent by mid-2024—momentum has cooled elsewhere. Mid-sized firms (20–250 employees) were flat to down, while only very small businesses (fewer than four employees) logged a modest uptick. Because BTOS runs every two weeks, this could prove to be a short-lived blip, but it’s the first clear sign of fatigue in the adoption curve.
Under the surface, the slowdown highlights practical limits to near-term ROI. Roughly 95 percent of surveyed companies reported no new revenue tied to AI. Labor dynamics aren’t helping either: recent research indicates AI exposure has disproportionately affected younger workers, correlating with a 13 percent drop in jobs for early-career roles while senior positions remain more insulated. And even when AI is deployed, human oversight is still required to catch errors, blunting cost-cutting promises and complicating rollouts. Adding to the skepticism, some next‑generation AI models have underwhelmed on benchmarks—another reminder that headline-grabbing demos don’t always translate into dependable business value.
Markets, however, are telling a different story. Major tech names continue to rally, buoyed by unrelenting demand for AI infrastructure. Even with occasional dips—like the pullback that followed commentary around sold-out AI GPUs—investors are still pricing in long-term growth. Yet analysts warn that if enterprise projects keep missing profitability targets, sentiment could cool quickly. In fact, a few companies have begun rehiring roles they previously cut, acknowledging that current AI tools aren’t fully replacing human work.
What to watch next is whether this dip marks a temporary pause or the start of a longer consolidation. If adoption doesn’t reaccelerate and most pilots remain unprofitable, the industry could face the kind of slowdown often described as an AI winter. More data over the coming months will be key to separating a short-term wobble from a structural shift in enterprise AI adoption.
