Adoption of artificial intelligence (AI) by large enterprises has dipped slightly, even as overall corporate use of the technology continues to grow, according to new data from the US Census Bureau.
Figures from the Bureau’s summer surveys show that uptake among companies with more than 250 employees fell from a peak of around 14% to 12%. Analysts say the modest decline reflects hesitancy among firms that have already invested heavily in AI infrastructure, from data centres to rollout support, but are yet to see clear returns.
Across the broader corporate landscape, however, adoption remains on an upward path. The Bureau’s biweekly survey of 1.2 million businesses found that 9.7% of respondents had used AI in the previous two weeks, up from 8.8% earlier in the year. Sentiment is also improving, with 13.7% of companies saying they expect to use AI in producing goods or services within the next six months. Even so, nearly two-thirds of respondents reported no plans to use the technology, highlighting that mainstream adoption is still in its early stages.
Industry leaders caution against interpreting the figures as a sign that large enterprises are abandoning AI. Sheila Flavell CBE, chief operating officer at FDM Group, said the dip underscores the importance of training and strategy in unlocking value. “AI can only deliver results when people know how to use it effectively,” she said. “Practical, hands-on training helps build confidence and allows employees to see how AI can support their roles.”
Security and governance concerns also remain a hurdle. Andy Ward, senior vice-president international at Absolute Security, noted that more than a third of chief information security officers have banned certain AI tools, such as DeepSeek, due to privacy and control risks. “AI can transform detection and response,” he said, “but without robust resilience strategies, real-time visibility and clear governance, it risks adding vulnerabilities rather than solving them.”
Some analysts argue the Census Bureau’s methodology may understate adoption rates, since it only measures AI use in the production of goods and services, excluding applications in marketing, administration and customer service. UBS has suggested that despite the slowdown among larger firms, overall AI uptake is still moving faster than the adoption of e-commerce in its first two decades.
However, questions remain over whether current levels of corporate investment are justified. Torsten Sløk, chief economist at Apollo Academy, said the slowdown points to caution among big companies, while NYU Stern associate professor Arpit Gupta suggested that “trillions in AI capex should probably be reconsidered.”
A recent report from MIT added to the uncertainty, finding that 95% of companies are struggling to generate meaningful financial returns from AI. Combined with signs of hesitation at the enterprise level, the findings have fuelled speculation that an AI bubble may be forming, with some firms pausing to reassess before committing to further expansion.
