Australia just ranked number one globally for Claude usage, adjusted for population. Not top five. Number one. And by a margin that should make you pause: Australians use Claude more than six times what our share of the global population would predict.

That stat comes from Anthropic’s new Cadences Economic Index report, published this month. It’s the first time they’ve published this kind of population-adjusted breakdown, and Australia sits at the top of it.

This isn’t out of nowhere. When Anthropic opened its Sydney office in early 2026, Australia was already tracking at four times the expected usage rate. In six months, that’s gone from 4x to 6x. The adoption curve here isn’t flattening.

The stat that actually matters

The headline number is interesting. But the finding that caught my attention is a different one.

The Cadences report breaks Claude usage into two categories. “Automation” is where the AI runs a task without the user in the loop. “Augmentation” is where the user stays involved: they’re working alongside the tool, not handing off to it.

In Australia, 54.5% of Claude usage is augmentation. The global average is 51.4%.

That’s a small gap in percentage terms. But it points to something specific about how Australians are actually using this tool. Not as a replacement. As a collaborator.

I’ve written before about the data risk hiding in your team’s AI habit. Staff are often using these tools well before any policy exists. The Cadences report adds a layer to that picture: they’re mostly using it to work alongside the tool, not to hand off whole tasks and walk away. That’s a different risk profile, and arguably a healthier one.

What people are actually producing

Anthropic found a meaningful output (something they call an “artifact”) in 93% of Claude conversations. That’s not chat. That’s work.

The breakdown: 20% of outputs are documents or reports. 9% are explanations. 7% are email drafts. The rest scatter across code, summaries, plans, and analyses.

This maps closely to what I’ve seen from small business owners using Claude seriously. They’re not asking it hypothetical questions. They’re using it to produce something they actually send to a client, present to staff, or put on their website. The tool has become part of the output process, not just a search alternative.

This is also why having a policy for what staff are doing with AI matters more than most business owners realise. If 93% of conversations produce a real output, then every unsupervised conversation is a real output. That includes the ones with confidential client details in the prompt.

The job risk question

The survey attached to the Cadences report asked about 9,700 Claude users about their own job security. Ten percent said they rated losing their own job as likely or very likely as a result of AI. More than a third expect their responsibilities to change significantly in the next 12 months.

Both numbers are worth sitting with. Ten percent isn’t small. That’s one in ten people who use this tool regularly, expecting it to take their job. But 90% don’t. And the augmentation-heavy usage pattern suggests most people are landing somewhere in the middle: AI changes what they do, but they’re still doing it.

For business owners, the implication is less about replacement and more about reconfiguration. What does a staff member who uses Claude for 40% of their document production do with the time that frees up? That question is mostly being answered ad hoc right now, without much guidance from the top.

Why Australia, though?

The report doesn’t answer this directly. A few things likely contribute.

English is our primary language, which removes the friction that still exists for non-English Claude users. We have high internet penetration and a relatively high concentration of knowledge workers as a share of the workforce. And Anthropic’s Sydney presence since early 2026 has probably accelerated enterprise adoption.

I’ve argued before that Australia’s AI ranking sounds good but the support infrastructure for small business hasn’t kept pace. The Cadences data sharpens that argument. We’re using this stuff more than anyone else, per capita. The question is whether small business owners are getting the same productivity gains as larger enterprises, or whether the adoption is concentrated in a subset of knowledge-work roles.

So what do you do with this?

If you’re a small business owner and your team isn’t using AI tools yet, you’re now the statistical outlier in your own country. That’s not a reason to panic, but it is a reason to get curious about what you’re missing.

If they are using it, the augmentation stat suggests most of them are doing it in a reasonably healthy way: staying in the loop, using Claude as a thinking partner. But it’s still happening without visibility into what’s being shared, what’s being produced, and whether it’s accurate.

One more thing worth flagging: the Cadences report covers usage patterns today. It doesn’t say anything about what happens when one provider’s tools become so embedded in your workflows that switching costs make you dependent on them. Given how quickly Australia has adopted Claude specifically, that concentration risk is worth keeping in mind. I’ve written about not putting all your AI eggs in one basket before, and this data makes that point a bit more urgent.

Australia is, apparently, the world’s most enthusiastic adopter of this particular AI tool. That’s genuinely interesting. Whether it translates into a competitive advantage for Australian businesses, or just means we’re taking on more dependency than we realise, probably depends on how deliberately owners and managers engage with it from here.

: News Deep Dives →

📖 More deep dives: News Deep Dives

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs

More hot takes: Hot Takes →