Canberra Is Arguing About Data Centres. Small Business Owners Are Still Copy-Pasting Into ChatGPT.

The Federal AI Inquiry wrapped four days of testimony this month. Google showed up. Microsoft showed up. OpenAI, Anthropic, Firmus, all there. The committee heard about sovereign data centres, frontier model licensing, copyright law, and the competitive threat from the United States and China.

What barely got a mention: the 2.5 million small businesses that still haven’t figured out where to start.

FinTech Australia made this point directly. The inquiry is too fixated on data centre sovereignty and big-model policy. The practical adoption problem, the one affecting Australian businesses right now, gets almost no floor time. RMIT’s framing was sharper: Australia does not have an AI awareness problem. Business owners know AI exists. The problem is implementation. They know the word. They don’t know what to do on Monday morning.

CPA Australia was blunter. Infrastructure access does not automatically produce adoption. “Not like build it and they will come,” was the framing. You can greenlight every data centre in the country and a bookkeeper in Ballarat is still going to spend twenty minutes copy-pasting client notes into ChatGPT because nobody told her there’s a better way.

The gap between what Canberra is debating and what small business owners actually need is not a slight misalignment. It is a different conversation happening in a different room.

The New York Fed published a small business survey last year. Small businesses that had adopted AI were more optimistic about hiring and revenue growth than those that hadn’t. That benefit only shows up for businesses that have actually started. The ones still in “I’ve heard about it” territory don’t get the benefit, and right now that’s most of them.

The inquiry witnesses aren’t wrong, exactly. Sovereign AI infrastructure matters. Frontier model licensing matters. These are legitimate policy questions. But they’re being treated as the primary problem when they’re actually a downstream concern. Fix the infrastructure, and a small business owner still has no idea which tool to use for their invoicing. The path from “Australia has sovereign compute” to “a cafe owner in Cairns automates her roster” has a lot of steps in the middle, and nobody in the inquiry room is mapping them.

The AI skills gap post covers the practical frame:

Does the tool do one specific thing? Can you check its output in thirty seconds? Does it fit into what you’re already doing?

Clear all three, try it for two weeks. That’s the whole framework. Not inspiring, but it works better than waiting for Canberra to sort out the licensing regime.

The inquiry will produce a report. The report will make recommendations. Some of those recommendations will eventually shape policy. That process takes years. Small businesses deciding whether to adopt AI are making that call this quarter.

There’s a version of this inquiry that spends more time on the real problem: why businesses know AI exists but can’t figure out where to apply it, what kinds of support actually move the needle, and the difference between awareness campaigns and implementation help. That version would be more useful to the 2.5 million businesses trying to work out their next step.

For more on where the federal AI discussion is heading, the full AI Inquiry recap has the full picture.

Sources

• FinTech Australia testimony, Federal AI Inquiry, October 2026
• RMIT testimony, Federal AI Inquiry, October 2026
• CPA Australia testimony, Federal AI Inquiry, October 2026
• ABS: Characteristics of Australian Business
• New York Fed: Small Business Credit Survey, Report on AI Adoption (2024)

More hot takes: Hot Takes →