Large enterprises could be wavering on AI adoption
AI adoption rates have dipped, but it's probably nothing to worry about for IT leaders
Has AI already peaked? New research points to a decline in AI adoption rates at large US companies.
Every two weeks, the US Census Bureau surveys 1.2 million American companies, asking a range of questions. Since 2023, that has included a question about whether or not a company made use of AI in the last two weeks.
While the general trend is upwards, there's been a decline in the use of AI for companies with more than 250 employees. That cohort is the group most likely to use AI, according to the bureau, peaking just shy of 14% and falling back down to 12% over summer surveys.
While this represents a small – and perhaps temporary – decline, it does suggest some degree of hesitation on the part of large enterprises, some of which are growing frustrated at poor returns on investment.
"The bottom line is that the biweekly Census data is starting to show a slowdown in AI adoption for large companies," noted Torsten Sløk, chief economist at Apollo Academy, in a post on the company's website.
In a post on social media, Arpit Gupta, an associate professor of finance at NYU Stern shared Apollo's chart and was asked what he considered the key takeaway.
Gupta noted that "trillions in AI cap ex should probably be reconsidered."
Sign up today and you will receive a free copy of our Future Focus 2025 report - the leading guidance on AI, cybersecurity and other IT challenges as per 700+ senior executives
Those results come as companies continue to spend huge sums on data centers and other infrastructure to support AI rollouts.
Similarly, a report from MIT last month revealed 95% of companies are struggling to deliver financial return, fueling claims of a pending AI ‘bubble’.
AI adoption rates remain strong overall
Across all companies, the picture is a bit different. The 1.2 million companies were asked whether they used any type of AI over the last two weeks, with 9.7% saying yes — an increase from the survey two weeks prior at 8.8%.
However, that result was a significant dip from previous editions, which were trending above 9%. At the end of 2024, the figure was 6.3%, suggesting a slow upward trend.
In the most recent survey, 83.8% said they had not used AI of any type over the last two weeks, with 6.5% of companies reporting they weren't sure.
Indeed, UBS analysis from June noted that growth was steadily increasing — though that was written before the most recent dip.
"This means AI adoption is likely to soon cross the 10% threshold that took US e-commerce 24 years to reach," UBS noted.
Questionable question
These AI usage rates may sound low given the amount of attention the technology has been given of late.
While that may be partly down to hype surrounding the nascent technology, another issue is the phrasing of the question asked by the Census Bureau: "In the last two weeks, did this business use Artificial Intelligence (AI) in producing goods or services?"
Plenty of businesses will be using AI for automating customer service or other administrative tasks, for marketing, or for project marketing, noted one analyst – tasks that respondents may not see as part of directly producing a good or service.
The survey also asks whether the company expects to be using AI in the next six months, again to make products or services. In the most recent poll, 13.7% of all respondents said yes, up from 9.2% at the end of last year.
However, nearly two-thirds still said they don't expect to be using AI within that time frame.
Make sure to follow ITPro on Google News to keep tabs on all our latest news, analysis, and reviews.
MORE FROM ITPRO
- AI adoption is finally driving ROI for B2B teams in the UK and EU
- Questions raised over AI’s impact as studies tout conflicting adoption outcomes
- A quarter of firms still don’t have a formal data strategy
Freelance journalist Nicole Kobie first started writing for ITPro in 2007, with bylines in New Scientist, Wired, PC Pro and many more.
Nicole the author of a book about the history of technology, The Long History of the Future.
-
Licensed mmWave: Opportunity or overhead?Industry Insights Ofcom’s latest mmWave auction unlocks major new capacity for 5G and FWA, offering a faster, more flexible complement to fiber - especially in dense urban areas
-
CISA issues alert as China-linked hackers exploit Brickstorm malware to target VMware serversNews Organizations, particularly in the critical infrastructure, government services, and facilities and IT sectors, need to be wary of Brickstorm
-
AWS CEO Matt Garman says AI agents will have 'as much impact on your business as the internet or cloud'News Garman told attendees at AWS re:Invent that AI agents represent a paradigm shift in the trajectory of AI and will finally unlock returns on investment for enterprises.
-
Westcon-Comstor partners with Fortanix to drive AI expertise in EMEANews The new agreement will help EMEA channel partners ramp up AI and multi-cloud capabilities
-
Microsoft quietly launches Fara-7B, a new 'agentic' small language model that lives on your PC — and it’s more powerful than GPT-4oNews The new Fara-7B model is designed to takeover your mouse and keyboard
-
Anthropic announces Claude Opus 4.5, the new AI coding frontrunnerNews The new frontier model is a leap forward for the firm across agentic tool use and resilience against attacks
-
Gartner says 40% of enterprises will experience ‘shadow AI’ breaches by 2030 — educating staff is the key to avoiding disasterNews Staff need to be educated on the risks of shadow AI to prevent costly breaches
-
Google blows away competition with powerful new Gemini 3 modelNews Gemini 3 is the hyperscaler’s most powerful model yet and state of the art on almost every AI benchmark going
-
Microsoft's new Agent 365 platform is a one-stop shop for deploying, securing, and keeping tabs on AI agentsNews The new platform looks to shore up visibility and security for enterprises using AI agents
-
Some of the most popular open weight AI models show ‘profound susceptibility’ to jailbreak techniquesNews Open weight AI models from Meta, OpenAI, Google, and Mistral all showed serious flaws
