The dramatic sell-off in NextDC reflects a broader loss of faith in the economics of data centre investing, according to a new analysis from the University of Melbourne. For two decades, data centres were seen as safe, lucrative real estate investments, the essential plumbing of the internet. But the rise of AI, with its insatiable demand for computing power, has launched a race to build ever larger facilities, and that race is now hitting roadblocks.
NextDC is not alone in facing headwinds. Data centre company Firmus abruptly cancelled its multibillion-dollar listing on the Australian stock exchange due to a lack of investor demand. London-based Nscale is struggling to attract buyers for a Wall Street listing. Shares in CoreWeave, which operates what it claims is the world's most powerful AI supercomputer, dropped 15 percent in a single day last week on financial strength concerns.
The analysis warns that the jitters reflect a harsh reality: while the hype around AI remains strong, the business models are coming under increasing scrutiny. Hundreds of billions of dollars have been spent globally on training models and building the data centres to house AI infrastructure, with the investment pipeline expected to reach some US$10 trillion (A$14.25 trillion) by 2032. Much of this spending has been funded by borrowing, and rising interest rates are now complicating the picture, making other investments more attractive.
A particular concern is the short lifespan of AI chips. Traditional data centre hardware could operate for decades with minimal maintenance, but AI chips need replacement every few years, driving a treadmill of multibillion-dollar upgrade expenses for power delivery and cooling. Even if AI revenues stabilise, there is no guarantee that data centre companies will be able to access the chips they need. Amazon recently admitted it is hoarding billions of AI chips without the capacity to install them, a sign of supply-chain anxiety.
Beyond the balance sheet, the analysis points to a loss of social licence for data centres amid local protests. Community concerns include persistent noise from energy transmission, visual impacts, localised heating, and environmental effects. Forecasts suggest that New South Wales data centres could consume up to 25 percent of Sydney's daily water demand within a decade. In Victoria, transmission company AusNet is reviewing requests from data centres for more than 10 gigawatts of connections, which could exceed the state's entire supply capacity.
The analysis suggests that to shore up investor confidence, AI investments should not just focus on building better models, but also models that are more efficient in terms of costs, energy demands and spatial requirements. If data centres can be designed to support the electricity grid by adding renewable capacity, then communities may begin to feel the benefits.
