Australia's AI adoption gap is bigger than the 12% headline suggests
ABS says 12% of Australian businesses use AI. The real story is 35% of large businesses against 11% of small ones, and the barrier isn't the technology.
The Australian Bureau of Statistics put out a number in June 2026 that got picked up everywhere: 12 per cent of Australian businesses reported using AI in their workplace in 2024-25, up from what the ABS describes as very low rates in 2021-22, before generative AI tools such as ChatGPT were available. Read as a headline, that sounds like a slow, even rollout across the economy. Read as a table, it is not that at all.
The same release breaks the 12 per cent down by size. Large businesses: 35 per cent, up from 9 per cent in 2021-22. Medium businesses: 22 per cent, up from 3 per cent. Small and micro businesses: 11 per cent. The national figure is an average sitting on top of a gap of more than three to one between the biggest firms and the smallest.
If your business has fewer than twenty people, the 12 per cent headline was never really about you.
Where the gap actually sits
This is not a story about small businesses refusing AI. The ABS data shows small firms that are already innovating adopt it at close to the rate of large ones: 19 per cent of innovation-active small businesses use AI, against 4 per cent of small businesses that aren’t innovating at all. The gap is not appetite. It is everything that sits between wanting to use AI and having a working version of it in the business.
A University of Queensland survey covering more than a thousand SMBs, reported in The Conversation in late 2025, sized that gap in more useful terms than a yes/no adoption count. About two-thirds of the firms surveyed had adopted AI “in some form.” But nearly 40 per cent of all respondents described their own use as still minimal. Just over a fifth had reached “moderate” adoption, using AI for something like demand forecasting or inventory management. Fewer than one in ten had embedded AI into anything more advanced, like fraud detection.
Two-thirds adoption and under 10 per cent doing anything substantial with it are both true at once. The first number is what gets quoted. The second is what determines whether the AI spend shows up in the P&L.
The barriers are specific, not vague
CPA Australia’s members-facing research on the topic put a shape on why. Surveying its base on AI barriers, more than a third of responses (34 per cent) named a lack of depth in understanding AI, its benefits, and its risks as the sticking point. One in five (20 per cent) named cost: subscription fees, pricing and licensing structures that don’t map cleanly onto a small operation. Smaller shares cited integration trouble getting AI to talk to existing systems (9 per cent), and data, privacy or ethical concerns (around 8 per cent and 6 per cent respectively).
None of that is really a technology gap. It’s knowledge, money, and time, in that order, and it should sound familiar: it’s the same trio Deloitte Access Economics landed on for CEDA when it looked at why smaller Australian firms lag on AI at all. Deloitte’s language for it was blunter still: “a lack of three things: time, knowledge and capital.”
The UTS Human Technology Institute’s SAAM research, drawn from SMEs already using AI rather than firms considering it, adds a workforce-side version of the same finding: 34 per cent of respondents pointed to gaps in initial knowledge about AI use, benefits and risk, and 20 per cent cited financial barriers. It’s the same shape from a different sample, which is the closest thing this kind of research gets to a second opinion.
There’s a geography layer on top. Research for the National AI Centre found regional Australian SMEs are 11 percentage points less likely to have implemented AI than their metro counterparts, with more than a quarter unaware of what AI could do for their business, against 19 per cent of metro firms. For a Perth-based business outside the CBD corridor, that’s not an abstract statistic.
What’s actually at stake if the gap doesn’t close
Deloitte’s modelling for its “AI edge for small business” report is the number worth sitting with. Just 5 per cent of the SMBs it surveyed were, in its terms, fully enabled to get the benefit AI can offer. Most were using it intermittently; a third weren’t using it at all. Its estimate: if just one in ten SMBs moved up one rung on that adoption ladder, from basic to intermediate use, or from intermediate to fully enabled, the effect on Australian GDP would be in the order of $44 billion a year.
At the firm level, the same report puts the profitability effect of moving from basic to intermediate maturity at roughly 45 per cent, and intermediate to fully enabled at around 111 per cent. Those are large, self-reported, vendor-adjacent figures and worth treating as directional rather than exact. But even discounted heavily, they describe a gap between “using AI” and “getting the value of AI” that is worth more than the adoption headline itself.
MYOB’s own transaction data, comparing AI-using small businesses against non-users, found the AI-using group growing 2.8 times faster. That’s a vendor measuring its own customer base, so it deserves the same discount. It points the same direction as everything above it, though: the ceiling on what AI does for a small business isn’t whether the business has tried it. It’s whether the business has gotten past the trying stage.
The gap is closeable, and it isn’t a technology problem
Every serious piece of research behind these numbers agrees on where the barrier actually sits, and it isn’t capability. It’s knowledge of what to use AI for and how, the time to work out where it fits into an existing operation, and enough certainty about the cost and the payoff to commit to it. That is a diagnosis problem before it is a technology problem: what in this specific business is actually worth automating, and what would it take.
That’s the gap a Workflow audit exists to close: not “should this business use AI,” which the data above already answers, but where, specifically, and at what return. For a business that wants to move past minimal use without guessing, that diagnosis, followed by a supported first step, is what AI Working Sessions are built to deliver. Start with a conversation.
Sources: Australian Bureau of Statistics, Business adoption of Artificial Intelligence accelerates in 2024-25 (media release, 25 June 2026) and Characteristics of Australian Business, 2024-25; Stan Karanasios, The Conversation (November 2025); CPA Australia, AI for SMEs: Overcoming cost and integration barriers, INTHEBLACK (May 2025); John O’Mahony for Deloitte Access Economics, CEDA opinion (November 2025) and Deloitte Access Economics, The AI Edge for SMBs, commissioned by Amazon (November 2025); UTS Human Technology Institute, In their words: perspectives and experiences of SMEs using AI (February 2025); Fifth Quadrant for the National AI Centre, Australian SMEs: AI adoption trends (January 2025); MYOB press release (April 2026).