Australia has spent the better part of two decades comfortable with the idea that someone else can make the things we use and, increasingly, think the thoughts our machines rely on. A new warning suggests that comfort is starting to look like a liability. Industry figures and economists say the country is drifting towards serious economic exposure if it continues to lean on imported goods and imported artificial intelligence at the same time, leaving domestic makers to wither and the broader economy hostage to decisions taken in Silicon Valley, Shenzhen or wherever the next supply shock happens to land.
The message, reported this week by 7News, ties two anxieties together that are usually discussed in separate rooms. One is the long, familiar decline of Australian manufacturing, where local producers are being undercut by cheap overseas imports and struggling simply to stay open. The other is the newer worry about sovereign AI, the question of whether a mid-sized economy can afford to run its industries, its government services and eventually its national security on models, chips and cloud platforms it neither owns nor controls. The argument being put forward is that these are not two problems but one, and that solving them apart is a recipe for solving neither.
Why the two problems are really one
The logic runs like this. Modern manufacturing is no longer about cheap labour bolting parts together. It is about automation, robotics, predictive maintenance, quality control by computer vision and supply chains optimised by software. All of that is increasingly powered by AI. A country that has hollowed out its factories has also, quietly, hollowed out the demand that would justify building AI capability at home. And a country that buys all its AI from abroad has no obvious way to rebuild advanced manufacturing, because the intelligence layer that makes a modern plant competitive sits on someone else’s servers, priced in someone else’s currency and subject to someone else’s export controls.
That interdependence is what has the warning’s authors reaching for a word as strong as chaos. If a geopolitical rupture, a pandemic-style shock or simply a commercial decision by a foreign vendor cut Australia off from either advanced components or advanced software, the economy would have limited ability to improvise a domestic replacement. The country learned a version of this lesson during the COVID years, when it could not reliably source masks, medicines or microchips. The concern now is that the next version of that vulnerability will be measured in tokens and GPUs rather than pallets of PPE.
The case for building at home
Advocates of a more muscular industry policy argue that Australia has most of the raw ingredients to do better. It has the lithium, the rare earths, the iron ore and, increasingly, the copper that the global electronics and energy transition depend on. It has cheap sun and wind that could, in principle, power the data centres AI runs on. It has strong universities and a respectable research base through bodies such as the CSIRO. What it has lacked, on this view, is the will to knit those advantages into finished products and into a domestic AI stack, rather than shipping the ore out and buying the intelligence back at a premium.
The prescription that tends to follow is a familiar one from the manufacturing debate: targeted procurement that favours local suppliers, incentives for firms that build advanced capability onshore, and public investment in the compute infrastructure that private companies find hard to fund alone. Supporters point to the Albanese government’s own Future Made in Australia agenda as evidence that the appetite exists, even if the AI dimension has so far been treated as an afterthought rather than the centrepiece.
The sceptics push back
Not everyone accepts the framing, and it would be dishonest to pretend the case is settled. Economists of a more orthodox bent warn that Australia has tried to prop up local manufacturing before, from cars to textiles, and mostly ended up subsidising industries that could not compete once the support was withdrawn. Their concern is that dressing the same protectionism in AI language does not change the underlying maths. A country of 27 million people, they argue, will never match the scale of the United States or China on frontier models or mass production, and money poured into chasing that parity is money not spent on the things Australia genuinely does well.
The more nuanced version of this critique does not reject sovereign capability outright. Instead it argues for being selective: pick the niches where Australia has a real edge, such as smaller specialised models, agriculture technology, mining automation and defence-adjacent applications, and buy the commodity layers from whoever makes them cheapest. On that reading, the goal is not to make everything at home but to make sure the country is never wholly dependent on any single foreign supplier for anything critical. It is a difference of degree rather than kind, but the degree matters enormously when the bill arrives.
What it means for Australia
For Australian businesses and workers, the debate is not abstract. Local manufacturers already operating on thin margins are the ones being squeezed by imports today, and they are also the firms least able to afford the AI tools that might let them compete. That is precisely the gap policy would need to close, through cheaper access to compute, skills programs and demand signals that reward buying Australian-built technology. For the wider economy, the stakes are about resilience: whether the country can keep its lights on, its hospitals running and its supply chains moving if the global system that currently supplies its goods and its software stops being reliable.
There is also a jobs dimension that cuts against the usual AI anxiety. If AI is going to reshape work regardless, the argument goes, Australia is far better off owning some of the value it creates than simply absorbing the disruption while the profits flow offshore. Building domestic capability is, in that sense, as much about capturing wages and tax revenue as it is about national security.
What’s next
The immediate question is whether the warning translates into policy or evaporates like so many manufacturing revival plans before it. Watch for how the federal government positions AI within its industry agenda over the coming year, whether the promised national AI framework carries any hard commitments on domestic compute and procurement, and whether the states, several of which are courting data centre investment, coordinate or compete. The uncomfortable truth is that both manufacturing and AI reward scale and patience, two things Australian politics has historically found difficult to sustain. The experts sounding the alarm are betting that the cost of continued drift is finally large enough to force the issue.
Sources: 7News
