Companies That Don’t Use AI: Why Most Businesses Still Skip It in 2026
Key takeaway: Most companies still don’t use AI. About 82% of US firms reported no AI use at the end of 2025, and only 20% of EU enterprises used any AI technology. The holdouts are not mainly technophobes: the biggest reasons are missing expertise, unclear rules, data leaks, and costs that don’t yet add up.
Companies that don’t use AI are the majority, not the exception. AI fills the headlines, so it is easy to assume every business already uses it. The data says otherwise: most businesses — especially small ones — have not adopted AI at all, and many that looked at it decided to wait. Here is how many there are, why they hold back, and what it costs them.
What counts as a company that doesn’t use AI?
“Not using AI” covers three situations: companies that never tried an AI tool, companies that tried one and stopped, and companies that actively ban AI tools at work. A firm that blocks public chatbots but runs AI inside its own software is not really AI-free — it has just drawn a line around where AI is allowed. If you are still deciding what AI means for your business, our plain-English explainer on how agentic AI differs from earlier forms of AI is a useful starting point.
How many companies don’t use AI?
In the United States, the Census Bureau’s Business Trends and Outlook Survey found that about 18% of firms used AI at the end of 2025, according to a Federal Reserve review of the data. That leaves roughly 82% of firms not using AI in any business function. About 21% of firms expected to start using AI within six months, so the non-user share is shrinking — slowly.
The gap is mostly about size. Weighted by employment rather than by firm count, the AI-use rate was 32%, so larger employers adopt AI far more often than small ones. Europe shows the same pattern: Eurostat’s 2025 survey found 20% of EU enterprises with ten or more employees used at least one AI technology, but only 17% of small firms did, against 55% of large firms.
Sectors differ sharply too. In the US, about 37% of information-sector firms used AI, along with roughly a third of professional-services firms and about 30% of finance firms. At the other end, only about 8% of accommodation and food-service firms did.
Why companies that don’t use AI stay on the sidelines
Eurostat asked EU firms that had considered AI but decided against it why they held back. About 7 in 10 cited a lack of relevant expertise, followed by unclear legal consequences and data-protection worries, each named by about half. Only about 1 in 5 said AI simply was not useful for their business. Most non-users looked, saw the gaps, and waited.
American enterprise leaders say much the same. In a Zapier survey of 532 enterprise leaders in September 2025, 78% said they were struggling to integrate AI with their existing systems, 45% named high vendor costs as a key barrier, 35% pointed to AI skill gaps, and 38% said they did not trust AI vendors’ security. Only 4% said they were actively resistant to adopting AI. The holdout is usually practical, not ideological.
Two famous cases show the caution is rational. In March 2023, engineers at Samsung pasted proprietary source code and a meeting transcript into ChatGPT within weeks of being allowed to use it, and Samsung temporarily restricted generative AI tools on company devices from May 2023. Banks including JPMorgan, Bank of America and Citigroup had already restricted ChatGPT earlier that year. And in June 2023, a New York federal judge fined two lawyers and their firm $5,000 after they filed a brief citing six court cases invented by ChatGPT. Accuracy and confidentiality are real reasons to pause — see our guide to responsible AI for risk decisions for how regulated firms handle this.
Companies that ban AI vs companies that never tried it
| Type of non-user | What they do | Main reason | Typical example |
|---|---|---|---|
| Never tried it | No AI tools in any workflow | No expertise, no clear use case, cost | Small firms in hospitality, retail, trades |
| Banned or restricted it | Blocks public AI tools on work devices | Data leaks, confidentiality, compliance | Samsung in 2023; several large banks |
| Tried it and stopped | Piloted AI, then shelved it | Poor integration, unclear ROI, bad data | Enterprises stuck between pilot and production |
The three groups need different advice. A company that never tried AI needs a first low-risk use case and basic training. One that banned AI tools needs approved, private alternatives. One that tried and stopped usually needs better data and integration before spending more on models. Governments face the same questions: our guide to AI agents in government explains the rules public agencies are setting.
What does it cost to stay AI-free?
Staying AI-free has no licence fee, but it is not free. Vendor prices were the top barrier for 45% of enterprise leaders in the Zapier survey, and integration, training and data clean-up sit on top of the subscription. Our breakdown of conversational AI consulting costs shows how quickly a pilot can run into five figures before it delivers anything.
The cost of waiting is harder to measure but just as real: competitors that automate quoting, support or reporting can answer faster and cheaper. Yet rushing in is no bargain either. BCG found in 2024 that 74% of companies had yet to show tangible value from AI, and McKinsey’s 2025 survey found 88% of organisations used AI in at least one function while nearly two-thirds had not begun scaling it. Paying for AI and getting nothing from it is the worst of both worlds, which is why the cheapest path is usually a small, measured pilot rather than a big purchase or a total refusal.
What a company needs before it starts using AI
A company moving from non-user to user needs four things before it buys a tool. First, one named use case with a measurable result, such as cutting first-response time in support. Second, someone on staff who understands the tool well enough to check its output — the expertise gap is the top barrier for a reason. Third, clean, organised data, since AI built on messy data produces unreliable answers. And fourth, a written rule on what staff may and may not paste into AI tools, so a Samsung-style leak does not happen on day one.
The risks of not using AI
The main risk of staying AI-free is competitive: rivals that automate routine work can serve customers faster and at lower cost, and the gap compounds over time. In sectors where AI adoption is already above a third of firms, such as information and professional services, a non-user is increasingly the exception.
But the risk runs the other way too. Adopting AI badly can leak confidential data, produce confident-sounding errors, and create legal exposure — as the $5,000 Mata v. Avianca sanction showed. The companies that navigate this best are not the fastest adopters or the proudest holdouts. They are the ones that treat AI as a controlled experiment: one use case, checked output, clear rules, and a willingness to stop what does not work.
Our take
Analysis: Not using AI in 2026 is usually a delay, not a strategy — the data shows most holdouts want to adopt and are blocked by skills, rules and integration, not by hostility to the technology. But the delay is rational while those gaps stay open: a leaked source code file or a fabricated citation costs more than a slow quarter. The winning move for a non-user is not to buy the biggest AI platform, and not to ban the topic either. It is to close one gap at a time — pick a use case, train one person, write the policy — and start small.
Frequently asked questions
What percentage of companies don’t use AI?
About 82% of US firms reported no AI use at the end of 2025, according to the Census Bureau’s Business Trends and Outlook Survey as reviewed by the Federal Reserve. In the EU, about 80% of enterprises with ten or more employees used no AI technology in 2025. Large firms adopt AI far more often than small ones.
Why do companies avoid AI?
The most common reasons are a lack of relevant expertise, unclear legal consequences, and data-protection worries. In Eurostat’s 2025 survey, about 7 in 10 firms that considered AI but held back named missing expertise. High vendor costs, trouble integrating AI with existing systems, and low trust in vendors’ security are also major barriers.
Which big companies banned AI tools?
Samsung temporarily restricted generative AI tools such as ChatGPT on company devices in May 2023 after employees pasted proprietary source code and a meeting transcript into the chatbot. JPMorgan, Bank of America, Citigroup and Goldman Sachs had already restricted ChatGPT earlier in 2023. Most of these were restrictions on public tools, not permanent bans on all AI.
Is it risky for a company not to use AI?
Yes, mainly in competitive terms: rivals that automate routine work can respond faster and cheaper. But adopting AI badly carries its own risks, including data leaks and inaccurate output. The safest path is a small pilot with checked output and a written policy on what staff may enter into AI tools.
How can a small business start using AI safely?
Pick one low-risk task, such as drafting routine emails or summarising internal notes. Use an approved tool, keep customer and confidential data out of public chatbots, train at least one person to check the output, and write down the rules before the wider team starts. Measure the result before expanding to a second task.
Sources
- Federal Reserve, “Monitoring AI Adoption in the U.S. Economy” (FEDS Notes, April 3, 2026)
- AI World, “The real obstacle to AI in European businesses is not the technology” (Eurostat 2025 survey data)
- Zapier, AI adoption survey of 532 enterprise leaders (October 28, 2025)
- Tech Monitor, “Samsung bans staff from using ChatGPT after data leak” (May 2, 2023)
- AI Vortex, “Mata v. Avianca: $5,000 Sanction for ChatGPT Fake Cases” (June 22, 2023)