There is a particular kind of nervousness that settles over a market when everyone is making money and nobody can quite explain why it should last. That mood now hangs over artificial intelligence, and it is the subject of a sharp piece of analysis in The Australian, which frames the current boom as a wager on a scale the technology sector has rarely attempted. The question it poses is deceptively simple. What happens when the music stops?
The framing borrows from an old line about dance floors and chairs, and it fits the moment. Global capital is flooding into AI infrastructure at a pace that has left even seasoned market watchers reaching for historical comparisons. The largest technology companies in the United States are collectively committing hundreds of billions of dollars a year to data centres, chips and power, on the expectation that demand for machine intelligence will keep climbing steeply enough to justify the outlay. So far, the spending has run well ahead of the revenue it is supposed to produce, and that gap is the heart of the gamble.
How the spending got here
The circularity is what unsettles the sceptics. Chipmakers invest in the model developers who buy their chips. Cloud providers sign enormous contracts with those same developers, then book the future revenue as validation of the build-out. Nvidia has become one of the most valuable companies on earth largely on the strength of orders from a handful of customers, several of which are themselves burning cash at a remarkable rate. When the buyers, sellers and financiers of a boom are so tightly interlinked, a wobble in one corner can travel fast.
None of this means the technology is a mirage. The tools work, they are being used by hundreds of millions of people, and the productivity case in specific tasks is real. The gamble is not about whether AI is useful. It is about whether it will be useful enough, soon enough, to repay the trillions being committed on its behalf. That is a different and much harder question, and it is the one the market has largely declined to ask out loud while the returns keep flowing.
Two ways of reading the same numbers
Bulls argue that the doubters are making the classic mistake of underestimating an exponential. Every previous general-purpose technology, from electricity to the internet, looked overbuilt at the moment of peak investment and undersupplied a decade later. On this view the fibre glut of the dot-com era was painful for early investors but laid the physical foundation for two decades of growth. The infrastructure being poured today, the argument runs, will look cheap in hindsight even if some of the companies building it do not survive.
The bears counter that the internet at least monetised quickly through advertising and commerce, whereas the current AI build-out is front-loading colossal fixed costs against revenue that remains thin and heavily subsidised. They point to the energy bill, the depreciation schedules on chips that age in a few years rather than decades, and the uncomfortable fact that much of the demand is being funded by the same players supplying it. Australian venture veteran Daniel Petre has warned locally of a “SaaSpocalypse” in which a wave of AI-native software wipes out incumbents and the valuations built on them, a reminder that even believers in the technology can be deeply wary of the prices attached to it.
Both readings can be partly right. It is entirely possible that AI transforms the economy and that a large slice of today’s investors lose money getting there. History is full of transformative technologies that ruined their earliest backers.
Why this matters in Australia
It would be comforting to treat all this as an American drama playing out on someone else’s balance sheet, but the exposure runs straight through the Australian economy. The nation’s superannuation pool, now well north of four trillion dollars, is heavily weighted towards global equities, and a meaningful chunk of recent returns has come from the handful of US technology giants driving the AI trade. When those stocks rise, Australian retirement balances rise with them. When they correct, the pain lands in the same accounts, whether or not the members ever typed a prompt into a chatbot.
The ASX has its own version of the story. Investors have chased local names with an AI angle, from healthcare and mining software to the information technology index itself, and sentiment here tends to import Wall Street’s mood wholesale. A sharp repricing offshore would not spare Australian tech shares simply because their fundamentals are steadier.
Then there is the physical build-out arriving on Australian soil. Data centre operators such as Firmus have attracted billions in offshore capital to construct AI-ready capacity, betting that demand for compute in the region will keep growing. Those projects carry real jobs, real power commitments and real property decisions in places like Sydney and Tasmania. If the global appetite for AI infrastructure cools, some of that pipeline could stall, leaving communities and grids that had been planning around it to adjust. The same money that makes the boom exciting is the money that makes a downturn consequential.
What to watch next
The near-term signals worth tracking are unglamorous but telling. Watch whether the big spenders start to soften their capital expenditure guidance, because the first sign of doubt usually shows up in a cautious sentence buried in an earnings call rather than a dramatic announcement. Watch the cost of financing the build-out, since much of it is increasingly being funded with debt rather than cash, and rising borrowing costs change the maths quickly. Watch, too, for the arrival of genuine enterprise revenue at scale, the thing that would turn the gamble into a sound investment rather than an act of faith.
For Australian investors and fund trustees the sensible response is neither panic nor blind enthusiasm. It is to understand how much of a portfolio’s recent performance rests on a single, correlated bet, and to size that exposure with clear eyes. The technology may well deliver everything its champions promise. The open question, as The Australian puts it, is what happens to everyone still on the dance floor if the tempo suddenly changes.
Sources: The Australian.
