Reserve Bank of Australia Assistant Governor (Economic) Sarah Hunter said at a UBS conference in Sydney on 9 November 2026 that AI investment was adding to demand and near-term price pressure, while its productivity benefits remained difficult to see. She also warned that disappointing investment returns could put AI-exposed asset prices at risk of a sharp reversal, Reuters reported.
Key points
- Hunter expects AI investment to add to inflation initially while supply is constrained.
- Domestic data centre investment has contributed to demand pressures as the RBA has raised rates five times this year, to 4.85%.
- Hunter sees limited evidence of a material productivity gain or large employment effects, alongside tentative signs of change in AI-exposed jobs.
- She warned that weaker-than-expected returns could bring sharp, potentially disorderly, asset-price adjustments.
Data centres and the RBA’s 4.85% rate
Hunter’s near-term concern is the demand created while AI capacity is being built. Investment in data centres and related activity requires resources before any resulting gains in output can ease price pressure. With supply constrained, she expects that additional demand to push inflation higher initially. Bloomberg reported that Hunter described rising investment in AI-related areas as a further source of demand while policymakers were raising interest rates to combat inflation.
The distinction between spending now and productive capacity later matters for a central bank confronting current prices. A boom in Australian data centre investment has contributed to demand pressure and is one reason the RBA has raised interest rates five times in 2026, taking them to an 18-year high of 4.85%, Reuters reported. Hunter’s assessment places AI investment among the forces bearing on the bank’s inflation task.
Hunter expects the balance to change as data centre capacity and related sectors expand. Greater supply would ease the bottlenecks she described. She said that, if AI produces broader productivity gains, economies could make more with the same inputs and the initial inflationary pressure from investment would probably dissipate. That later effect depends on gains that Hunter said have not yet appeared materially in aggregate productivity.
The RBA’s governor had also warned that AI investment was adding to inflation before productivity gains. Hunter’s account supplies a sequence for that concern: demand rises as capacity is built, supply catches up as facilities and related sectors expand, and any broader productivity benefit comes later. The timing leaves the bank dealing with investment’s immediate effect on demand rather than assuming a future benefit has already offset it.
Hunter sees tentative labour-market changes
Hunter said there was limited evidence that AI had materially increased productivity or had large effects on employment. Her assessment separates the technology’s aggregate economic results from the scale of investment directed towards it. The RBA can observe the demand from construction and related spending now, while the productivity improvement that might eventually ease inflation remains conditional in Hunter’s account.
She did, however, identify tentative signs of adjustment within the labour market. She said job-ad posting and employment outcomes were weaker in occupations more exposed to AI, while growth was stronger in areas less affected by the technology. Hunter drew particular attention to younger workers. These are narrower observations than a large effect on employment across the economy, and she presented them as tentative.
That difference matters to the order in which the effects arrive. Hunter described investment demand as an immediate source of price pressure, but treated both broad productivity gains and changes to employment as less evident. The pattern allows for change in particular occupations without treating those outcomes as proof of an economy-wide shift. It also leaves her inflation assessment resting on present spending and supply constraints.
Hunter’s account of a possible later productivity gain has a condition attached: AI must enable more production from the same inputs. Expanding data centre capacity could relieve bottlenecks in the meantime, she said. The two processes have different roles in her assessment — one eases a constraint created during the build-out, while the other would determine whether the wider economy ultimately produces more efficiently.
AI returns and asset-price risk
Hunter also raised a financial stability concern tied to the expected returns on AI investment. If those returns fall short, she warned, some asset prices could be sharply repriced or adjust in a disorderly way. This is a conditional risk, distinct from her description of investment already adding to demand. The spending can put pressure on prices in the near term even if the eventual return remains uncertain.
The RBA had previously raised a related concern about AI debt and financial exposures. Hunter’s warning at the Sydney conference focused on the possibility that asset prices exposed to the investment boom would have to adjust if expected returns failed to arrive. She did not need to predict that outcome to identify the risk: its trigger, in her account, is a shortfall against expectations.
For now, Hunter’s two concerns operate on different clocks. Building AI capacity adds demand while the economy is supply constrained. A later expansion of capacity could ease bottlenecks, and broader productivity gains could dissipate the initial inflation pressure if they materialise. The potential asset-price adjustment instead depends on whether investment delivers the returns being expected of it.
Hunter said weaker job advertisements and employment outcomes in more AI-exposed occupations were most notable alongside stronger growth in less-affected areas, particularly for younger workers. She characterised those labour-market differences as tentative, while warning that AI-exposed asset prices could face sharp repricing, or even disorderly adjustment, if investment returns disappoint.
