Andrew Leigh has a new piece in Startup Daily, and one number in it deserves more attention than it’s getting: only 7% of Australian small and medium businesses are making broad use of AI.

That might sound low. It is low. But here’s the thing that makes it interesting: earlier research from NAB put AI usage among SMEs at 42%, with another 14% planning to start. So which is it?

Both, actually. The figures measure different things. Using AI once in a while, drafting a social post, asking ChatGPT a question, counts as “using AI.” Weaving it into how you actually run the business, regularly, across multiple functions, that’s “broad use.” The gap between 42% and 7% is the gap between dabbling and adopting.

Leigh’s argument is that the gap matters enormously. He calls it the diffusion dividend.

The diffusion dividend

The basic idea: economic history is a history of spread, not invention. Electricity created value when factories reorganised around it. The internet created value when it changed commerce, education, and daily life. The invention was step one. The spread was where the productivity gains actually lived.

Leigh uses zero-till farming as his example, an agricultural technique that took 40 years to reach 80-90% adoption in Australian winter cereal regions. Nobody was holding a launch event. Farmers tested it, adapted it, shared it with other farmers. Local knowledge did what a brochure couldn’t.

His point is that AI will follow the same curve. The question is how long Australia takes to get there, and whether small businesses get left behind in the meantime.

What the 2026 Budget has to do with it

Leigh ties the adoption gap directly to two Budget measures.

First, the $20,000 instant asset write-off made permanent. This matters because investment in AI tools, software subscriptions, hardware, training, is exactly the kind of spending that benefits from immediate deductibility. A small business owner who can write off a new tool this financial year is more likely to actually buy it.

Second, scrapping non-compete clauses for workers earning under $183,000. This one is less obvious but makes sense in context: when workers can move more freely between employers, they carry knowledge and practices with them. Better employers attract talent. Worse employers have to lift their game. Ideas spread through people.

The risk isn’t automation. It’s falling behind.

There’s a piece of French firm-level research Leigh cites that’s worth sitting with. Companies in France that adopted AI are hiring more and selling more, including in roles that were supposed to be most at risk from automation.

His read: the biggest employment risk isn’t working in a job that AI might replace. It’s working in a business that doesn’t adopt AI at all, while competitors who do pull further ahead.

For an Australian small business owner, that reframe is worth taking seriously. The question isn’t “will AI take my job?” It’s “will my competitors use AI to do what I do, faster and cheaper, while I’m still figuring out whether to try it?”

What “broad use” actually looks like

Moving from 7% to something higher doesn’t mean overhauling your entire business. It means AI stops being the thing you try occasionally and becomes part of how you work.

In practice, for most small businesses, that looks like:

Using AI for quotes, customer updates, and follow-up emails, not just when you remember to
Running your rostering, invoicing, or stock management through tools with AI built in
Using AI to prepare for conversations (supplier meetings, difficult client calls, grant applications) rather than going in cold
Having someone in the business whose job includes figuring out where AI saves time

None of that is radical. It’s just consistent.

Leigh’s speech from last week, which we covered in detail, made the case for “boring AI” being the most valuable AI for small businesses. This op-ed is the same argument through a policy lens: adoption is an infrastructure problem, not just an individual business decision.

The $20k write-off helps with cost. The non-compete changes help with knowledge flow. But the main thing standing between 7% and something closer to 70% is small business owners deciding that broad use is worth the effort.

Given the alternative, it probably is.

📬 Get the SmallBizAI.au Weekly

Practical AI news for Australian small businesses — every week, free.

Andrew Leigh is Australia’s Assistant Minister for Competition. His op-ed was published in Startup Daily on 16 May 2026. For more on his earlier speech on AI and small business, see our full analysis from last week.

📬 The SmallBizAI Brief

One practical AI tip for Australian small business. Every Tuesday. Free.

Join business owners getting smarter about AI every week.

Subscribe free →