Reserve Bank warns AI boom will lift inflation before it lifts productivity

Story So Far

Australia's central bank has put a number on the artificial intelligence boom and a warning alongside it. Speaking at the UBS Australasia Conference in Sydney, Reserve Bank Assistant Governor Sarah Hunter said AI investment is already lifting demand and adding upward pressure on inflation, while there is still limited evidence it has lifted productivity across the economy.

The scale of the build-out is the core of her argument. The bank's estimates for realised data-centre investment have risen tenfold from a very low base over the past three years, and Hunter said investment in data centres and utility networks is already adding directly to demand, including in Australia's major trading partners. She put the expected data-centre spend at about 2 per cent of GDP a year, with $115 billion invested by 2028.

On productivity, Hunter was blunt about how little is proven. Estimates of AI's effect on annual productivity growth over the next decade range from just 0.1 per cent to more than 2 per cent, and the bank's own medium-term assumption for productivity growth is 0.7 per cent a year, unchanged since last August. She said almost no firms in the bank's liaison program now report zero AI use, down from around a third in 2025, yet that adoption has not shown up in aggregate output.

The speech and the question-and-answer session that followed it are published on the Reserve Bank's own site, and the reporting of them lines up. Reuters, the bank's transcript and independent summaries all carry the same central claims: near-term upward pressure on inflation, a tenfold rise in data-centre investment, limited evidence of a productivity lift, and a warning that disappointing returns could expose equities and corporate debt to sharp repricing. The cash rate figure of 4.85 per cent, an 18-year high after rises in September and November, appears in secondary reporting rather than in the transcript reviewed. Hunter also said the bank sees only tentative signs of weaker hiring in AI-exposed roles, with stronger growth in less-exposed areas.

Hunter drew the parallel with the internet rollout of the late 1990s, arguing that general-purpose technologies take years to show up in productivity and that the long-run effect of AI on inflation is likely to be broadly neutral once supply catches up with demand. In the near term, she said, the positive shock to aggregate demand set against the economy's current supply capacity will put some upward pressure on inflation.

The message lands on households already carrying five rate rises this year, and it frames the bank's dilemma plainly: the spending is real and rising, the payoff is not yet visible, and the market repricing risk grows if the returns disappoint. The Reserve Bank's next interest-rate decision comes at its December board meeting.