Artificial Intelligence and the Next Chapter for Australian Commercial Real Estate

Ask a room of executives what artificial intelligence means for employment and the answers cluster at the extremes: mass displacement on one side, a surge in productivity on the other. Both are likely wrong, and the gap between them is where the economic story of the next decade will play out. To understand this range of possible outcomes, Cushman & Wakefield has engineered the AI Barometer to track the adoption of AI in Australia, developed four possible scenarios of how AI may diffuse across the Australian economy over the next decade and modelled the implications of each of these on commercial real estate markets. 

 

Breakthroughs, bubbles and bottlenecks

The first is our baseline forecast, a gradual, at times bumpy adoption path for AI which provides a modest uplift to productivity through 2035. The second examines the consequences of an AI investment bust, where AI is still transformational in the long-term, but in the short-term AI companies cannot generate sufficient revenues to cover the costs of the required infrastructure build-out, data-centres, power etc., to support adoption. The result is a financial market disruption originating in the US, and transmitted to Australia first through credit markets then real economic impacts. 

The third and fourth scenarios share a premise of super-productive AI but split on what it does to workers. There is the upside case: AI complements rather than replaces workers, boosting productivity and growth in a manner reminiscent of previous technology booms. Finally, the scenario in which AI substitutes labour, where AI displaces cognitive labour faster than the economy can reabsorb it, and the productivity gains from AI accrue to capital. 

These scenarios result in some degree of labour force disruption, where some jobs are displaced or augmented and new jobs and fields of work are created. Crucially, the difference between these two scenarios rests on the Jevons paradox: when technology makes something cheaper, demand for it often rises rather than falls. For example, if AI makes legal services or healthcare cheaper, we may consume more of it rather than less, spurring employment growth: AI would reshape the labour force rather than reduce it. This nuance has been lost in most of the prevailing debates around AI, and given that Jevons has held following the introduction of every general purpose technology, we place a significantly higher probability on the complement scenario occurring than the substitute scenario; but both are modelled to understand the magnitude of the upside and downside risks.

Built from the bottom up

Rather than applying a single national assumption, our approach is built from the bottom up. Each scenario is modelled at the industry level: 19 industries, each with its own exposure to AI based on the tasks its workforce actually performs, then broken down to the city and submarket level based on the distribution of labour across each city. As a result each scenario can produce different results across Australia’s cities and submarkets. 

Shops, sheds and beds

For most CRE sectors, AI's impact arrives indirectly, through household income and population rather than technology itself. Retail tracks the consumer: it performs well where productivity gains flow through to wages, and suffers most where income shifts from households to capital. The Living sectors are the most insulated, anchored by demographics and chronic undersupply rather than the technology cycle. Logistics and Industrial has the highest floor across every scenario; both directly as highly productive AI would increase the share of online spending and indirectly as AI-enabled supply chains raise the value of well-located, automated facilities.

The white-collar wildcard

Office is where the scenarios diverge most sharply, because white collar employment is precisely the type of work which is most exposed to AI. But there are other mechanisms at work apart from direct office employment that affect demand for office floorspace. In the complement scenario, office net absorption is above the baseline, but the demand from additional workers is muted slightly by changing space demands and office attendance that an super-productive AI enabled workforce would entail. 

The substitution scenario is characterised by extreme bifurcation in office markets – profits accumulate to a small group of companies who compete for the most premium space – increasing the gap in rents for the best vs second best space in the market. The bust scenario has one of the most interesting implications for the office market: the downturn temporarily raises vacancy, but also freezes development for the foreseeable future. When the recovery comes supply isn’t elastic enough offset the boom in demand – in this scenario we forecast the prime vacancy rate in the Sydney CBD to fall well below the baseline 5% forecast by the early 2030s, even though the number of office workers employed is below our baseline forecast. 

Six cities, six stories

Geography matters too. The distribution of the workforces of Sydney and Melbourne across industries means that these cities carry the highest exposure and the highest upside, they are home to both the workers most affected and the firms best placed to capture the gains. Canberra is the most defensive market in every scenario, anchored by counter-cyclical government demand, while Adelaide's defence pipeline provides decade-long visibility largely independent of the AI cycle. Brisbane's Olympic construction pipeline provides a demand floor across even the downside scenarios - restructured in a credit crunch, but not derailed. Perth has relatively lower direct task exposure as AI's energy and hardware appetite supports demand for LNG and critical minerals.

AI will not deliver a single, uniform shock to Australian property; it will deliver different outcomes by city, sector and asset grade. In every scenario we model, though, one finding holds: the premium on quality grows. The buildings that help firms attract the people who work alongside AI will outperform, whichever path the technology takes.

If you would like more information or to dive deeper into the data contact us for a personal briefing.