Most Australian small businesses have already tried AI. Around two-thirds use it in some form, according to a Deloitte Access Economics report from November 2025. The harder truth sits underneath that headline: only 5% of those users are set up to get real value from it. Adoption is wide and shallow.
This article lays out what the data actually shows. How many businesses use AI, why most hesitate to go further, and what separates the handful seeing returns from everyone else.
AI adoption is the process by which a business moves from awareness of a technology to dependable, value-producing use of it. Adoption is not a single yes-or-no event. It is a ladder, from a staff member quietly using a chatbot, to tools wired into daily workflows, to systems the business trusts to run parts of its operation.
How many Australian small businesses use AI
The numbers depend on how you define use, which is why the headline figures move around.
The Deloitte Access Economics report, drawn from a survey of more than 1,000 Australian small and medium businesses, found two-thirds are using AI in some way. The Australian Government's National AI Centre, which tracks a tighter measure through its SME AI Pulse, puts current SME adoption in the mid-40s, reported at 44% in February 2026. At the individual level, the KPMG and University of Melbourne global study found half of Australians use AI regularly.
So the AI adoption rate in Australia is real and substantial, somewhere between 44% and two-thirds of small businesses depending on the threshold. The gap between those figures is the story. A business where one employee pastes the odd email into a chatbot counts as an adopter on a loose measure and a non-adopter on a strict one.
Deloitte's framing is the most useful here. It treats adoption as a ladder with rungs: basic, intermediate, and enabled. Two-thirds are on the ladder. Only 5% have reached the top. The distance between trying AI and depending on it is where most small businesses are stuck.
The cost of stalling on the second rung
The reason this matters is money, and the figures are not small.
Deloitte estimates that if just one in ten Australian SMBs advanced a single rung on the adoption ladder, $44 billion could be added to the country's GDP each year. That is not a forecast for full transformation across the economy. It is the result of a modest, plausible step up by a fraction of businesses.
At the level of an individual firm, the same report found a typical small business moving from basic to intermediate AI use could see a 45% increase in profitability. Moving from intermediate to enabled use could see a 111% increase. The returns compound as a business climbs, which means the businesses doing the least with AI are leaving the most on the table.
We see this pattern in our own work with small businesses. The first experiment with a tool is easy and the value is obvious within a week. The second step, building AI into how the business actually runs, is where momentum dies. Owners are busy, the tools feel risky, and nobody is sure who is responsible when the output is wrong. So the business camps on the second rung, and the gain stays small.
Why most small businesses hesitate: the trust barrier
The standard explanation for slow adoption is cost or technical skill. The Australian data points somewhere else: trust.
The National AI Centre found that among SMEs not adopting AI, trust is the leading barrier, cited by 65%. That covers distrust of AI decision-making and a preference to keep a person in control. Relevance comes second at 54%, businesses that believe AI does not apply to what they do. Cost and skills sit further down at 20%. Money is a factor, but it is not the main one.
The broader population data agrees. The KPMG and University of Melbourne study, which surveyed 48,340 people across 47 countries, found that while half of Australians use AI regularly, only 36% are willing to trust it, and 78% are concerned about negative outcomes. The work was led by Professor Nicole Gillespie and Dr Steve Lockey of Melbourne Business School. Australians use AI more than they trust it, and that gap is wider here than in many comparable countries.
This reframes the barriers to AI adoption. The instinct is to treat hesitation as ignorance to be corrected with more features and louder marketing. The data says the opposite. People have used these tools, watched them produce confident nonsense, and drawn a sensible conclusion: do not hand over anything that matters without checking. A hallucination, where a model states something false as fact, is not a rare glitch. It is a known property of the technology. Caution is the correct response, not a problem.
The skills gap is real underneath this. Deloitte found only about 10% of SMB workforces have advanced AI skills, with over half at basic or novice familiarity. But skills follow trust. People learn tools they believe are worth learning. Build the trust and the skills follow; ignore it and no amount of training sticks.
What the 5% do differently
The businesses getting value from AI are not the ones with the biggest budgets or the most tools. They share a few habits.
They start with a specific problem, not a product. Instead of asking what AI can do, they ask which task costs them the most time each week, then test whether AI helps with that one thing. This is the difference between adoption and what we have written about elsewhere as systems thinking for small business: improving the whole operation rather than bolting on a tool and hoping.
They fix their data first. Deloitte names business systems and data quality as a leading constraint. AI built on messy records produces messy answers. The enabled businesses cleaned up where their information lives before expecting a tool to read it. For anything involving a model answering questions from internal documents, the quality of those documents decides the quality of the output, which is the whole premise behind retrieval-augmented generation.
They keep a person accountable. The enabled businesses do not remove human judgement; they place it deliberately. A draft gets reviewed before it goes out. A recommendation gets checked before it is acted on. This is the human in the loop principle, and it is the single practice that lets a cautious owner expand AI use without lying awake about it.
They treat output as a first draft, not a finished product. The 5% understand what AI is good at, generating volume and speed, and what it is not, judgement and accountability. We have written separately on what AI can't do, because knowing the limits is what makes the strengths usable.
The Australian angle: a large sector with a lot to gain
The scale of what is at stake is easy to underestimate.
Small businesses make up 97.3% of all Australian businesses, 2,656,469 of them as of June 2025, according to the Australian Small Business and Family Enterprise Ombudsman citing ABS data. Most are very small: 64% are non-employing sole traders, 25% employ between one and four people, and 9% employ between five and nineteen. Small businesses also employ over five million people, around 39% of the private-sector workforce.
A few things follow. The Australian economy is unusually exposed to whether small businesses adopt AI well, because so much of it is small businesses. A productivity step among firms with one to four staff matters more here than in economies dominated by large enterprise. The $44 billion figure is large precisely because the base is so broad.
The composition also shapes what good adoption looks like. A sole trader does not need an enterprise AI platform. They need one or two tools that save real hours, set up by someone who understands their work. The trust barrier is sharper for them, not softer: a sole trader carries every error personally, so the bar for handing work to a machine is higher. Adoption that ignores this fails. Adoption that respects it, starting small, keeping the owner in control, and proving value before expanding, is what moves a business up the ladder.
Adoption is also uneven by sector. National AI Centre tracking shows adoption is led by sectors such as health, education and services, while construction and agriculture lag well behind. The lagging sectors are not less capable. They have had fewer relevant tools and clearer reasons for caution, which means the relevance barrier, not just trust, is doing real work there too.
The Enki Approach
We treat the trust barrier as the starting point, not an obstacle to talk past. Most small businesses we work with have already tried AI and come away unsure. Our job is to turn that uncertainty into dependable use.
That means starting with one costly problem rather than a tool catalogue, fixing the data the tool will rely on, and keeping a person accountable for anything that matters. It means being honest when AI is not the answer. The goal is not to get a business onto the ladder. Two-thirds are already there. The goal is to move it up a rung, where the Deloitte figures suggest the return is real, and to do it in a way the owner trusts enough to keep going. Practical adoption, with judgement kept in the room, is how a small business captures the upside without inheriting the risk.
Frequently asked questions
### What is the AI adoption rate among Australian small businesses? It sits between 44% and roughly two-thirds, depending on the measure. The Australian Government's National AI Centre reported SME adoption at 44% in February 2026 on a tighter definition, while a Deloitte Access Economics survey of more than 1,000 SMBs found two-thirds using AI in some form. The gap reflects how loosely or strictly "use" is defined.
### What is the biggest barrier to AI adoption for small business? Trust, not cost. The National AI Centre found 65% of non-adopting SMEs cite distrust of AI decision-making or a preference to keep humans in control. Relevance is second at 54%, and cost and skills sit at 20%. The KPMG and University of Melbourne study confirms the pattern: only 36% of Australians are willing to trust AI even though half use it regularly.
### How much could AI add to the Australian economy? Deloitte Access Economics estimates $44 billion could be added to Australia's GDP each year if just one in ten SMBs advanced a single rung on the AI adoption ladder. At the firm level, moving from basic to intermediate use is associated with a 45% lift in profitability, and intermediate to enabled with 111%.
### Why do most small businesses fail to get value from AI? Because adoption is wide but shallow. Deloitte found two-thirds of SMBs use AI but only 5% are "fully enabled" to capture its benefits. The businesses that succeed start with a specific costly problem, clean up their data first, keep a person accountable for output, and treat AI results as a first draft rather than a finished answer.
### Do small businesses need expensive tools to adopt AI? No. Cost ranks low among the barriers, cited by around 20% of non-adopters. Most small businesses, and 64% of Australian businesses are sole traders, need one or two well-chosen tools that save real hours, set up to fit their existing work. The constraint is usually knowing where to start and trusting the output, not the price of the software.