For much of the past two years, the story Australia told itself about artificial intelligence was one of relentless upside. Adopt fast, the argument went, and productivity would lift, wages would follow and the country would finally shake off a decade of stagnant output. That optimism is now colliding with a more sober mood in boardrooms, and a fresh analysis in the Australian Financial Review captures the shift with unusual bluntness.
Writing under the headline Australia’s AI gamble: Who wins, who loses, who leaves, the AFR’s James Thomson and Ronald Mizen argue that chief executives are quietly losing patience with the giant technology vendors selling the AI dream. At the same time, they suggest, the federal government is recalibrating its own approach, and the Prime Minister may be overplaying a hand that looks stronger in a speech than it does on a balance sheet.
Context: from hype cycle to reckoning
The frustration did not appear overnight. Australian businesses have spent freely on licences, pilots and consultants since generative AI went mainstream, often on the promise that measurable returns were just around the corner. In many cases those returns have proved slippery. Costs are easy to see on the invoice while benefits show up unevenly, buried in workflows that are hard to quantify. The result is a growing gap between the confidence of the sales pitch and the evidence on the ground, and it is that gap the AFR analysis puts under the microscope.
Boards are now asking harder questions. Where is the productivity, and can it be measured rather than merely asserted? Why are the recurring bills climbing even as usage patterns settle? And who ultimately captures the value created, the Australian firm doing the adopting or the offshore platform providing the model? These are not abstract concerns. They cut to whether the country is genuinely building an AI economy or simply renting one from a handful of American labs.
The news: a policy pivot and a patience problem
The central thread of the AFR piece is that two forces are moving at once. On one side, corporate Australia’s enthusiasm is cooling into scepticism as the vendor promises fail to fully materialise. On the other, Canberra is adjusting its policy stance after a period in which the government leaned heavily into an adoption-first message. The authors frame this as a moment of testing, where the winners, losers and those who walk away from the whole exercise are starting to become visible.
Prime Minister Anthony Albanese has made AI a signature theme, tying it to the broader productivity agenda that dominated last year’s economic debate. The AFR analysis questions whether the rhetoric has run ahead of the delivery, and whether a government keen to be seen backing the future risks writing cheques, political and otherwise, that the technology cannot yet cash. That is a pointed suggestion in a period when the productivity dividend from AI remains more forecast than fact.
Two views: believers and doubters
There are, broadly, two camps in this debate, and both have a case. The optimists point out that transformative technologies almost always disappoint in the short run and overdeliver in the long run. Electricity, the internet and cloud computing all went through a trough of scepticism before reshaping entire industries. On this reading, the current impatience is a natural phase, and the businesses that pull back now will regret it when the tooling matures and the workflows are rebuilt around it. Walking away, they argue, is the real risk.
The doubters counter that not every hyped technology becomes electricity, and that the discipline of demanding proof is exactly what has been missing. They note that the economics of large language models still favour the platform owners, that switching costs are rising as firms embed vendor tools deep into their operations, and that a strategy built on other people’s models leaves Australia exposed. The sceptical CEOs described in the AFR piece are not Luddites. They are managers who have been burned by pilots that never scaled and who now want returns before they commit further capital.
The Australian stakes
For Australia specifically, the stakes go beyond individual balance sheets. The country has framed AI adoption as a national productivity project, one meant to help lift the sluggish output growth that has troubled economists and the Reserve Bank alike. If the corporate appetite sours before the benefits land, the political narrative around AI as a growth engine becomes much harder to sustain, and the government’s own credibility on the issue takes a hit.
There is also the sovereignty question that has run through so much of the local debate this year, from calls for a genuine sovereign AI capability to unease about hosting rather than building the technology. If the value of AI flows disproportionately to offshore platforms, then every dollar Australian firms spend deepens a dependency rather than building domestic capacity. That is the uncomfortable subtext of the “who wins, who loses” framing. It is entirely possible for Australian businesses to adopt AI enthusiastically and still end up on the losing side of the ledger, with the gains banked in California.
The skills dimension sharpens the point further. Without the workforce capability to deploy these tools well, adoption becomes a cost centre rather than a capability. The businesses that win will be those with the internal expertise to separate genuine value from vendor theatre, and those able to negotiate hard on price and terms. Smaller firms and the public sector, with thinner technical benches, are the most likely to overpay and underdeliver.
What’s next
The immediate test is whether the government’s policy recalibration produces something businesses can actually use: clearer measurement of AI’s productivity impact, stronger support for local capability, and guardrails that build confidence rather than merely adding compliance. If Canberra can shift the conversation from adoption for its own sake to measurable, defensible value, it may reset the terms of the debate before disillusionment hardens.
For corporate leaders, the coming reporting seasons will force the issue. Investors are already nervous about stretched AI valuations globally, and executives who have spent heavily will be asked to show what they bought. Expect more firms to renegotiate contracts, consolidate their tooling and, in some cases, quietly wind back projects that never justified their cost. The AFR’s framing is a useful warning that Australia’s AI bet is genuinely a bet, with real losers as well as winners, and that the country has not yet settled which it will be.
Sources: Australian Financial Review (Technology).
