The pressing existential challenges of our era—climate change and artificial intelligence—are converging as evidenced by the rapid expansion of data centers in Australia and globally.
These developments have generated considerable discourse, with some expressing admiration for the potential advancements they could bring to humanity, while others voice concerns regarding their effects on the climate, inflation, employment, and housing costs.
Moreover, the ramifications of artificial intelligence itself—which can be both exhilarating and alarming—serve as a key factor driving the surge in data center construction.
Currently, there are over 10,000 operational data centers worldwide, with projections indicating this figure could increase by 3.5 times, costing approximately $7 trillion. To put this into context, this amount exceeds 5% of the global annual GDP, highlighting the scale of investment involved. While the United States is home to the majority of these facilities, Australia is witnessing a notable rise in activity, with 286 existing or planned centers. Prominent AI firms, such as Anthropic, are also eyeing Australia as a viable location for training their models.
The implications of this data center investment boom are significant from economic, environmental, and social perspectives. However, governments both locally and internationally have largely adopted a hands-off stance, possibly due to a fear of missing out on perceived benefits or a reluctance to challenge influential tech leaders.
Proponents of data centers often label them as “infrastructure,” a term that conveys necessity. Yet, unlike traditional “hard” infrastructure (like roads or utilities) or “soft” infrastructure (such as healthcare and education), it remains unclear who genuinely benefits from these investments and how. If data centers are to be classified as infrastructure, they should undergo rigorous scrutiny to evaluate whether their advantages outweigh the associated costs, akin to any other major project.
There is no denying that AI holds the potential to enhance human life—not just in trivial tasks, but in meaningful applications. For instance, in Shanghai, AI is easing traffic congestion, while globally, it is enhancing the accuracy and speed of medical imaging diagnostics and optimizing energy grids to prevent blackouts. The prospective economic and social advantages are vast, but these must be weighed against their potential drawbacks.
These challenges are considerable. The Australian Prudential Regulation Authority has alerted banks to the escalating cybersecurity risks linked to AI, recommending, somewhat ironically, that AI tools be employed to mitigate these threats.
It is widely recognized that data centers consume substantial amounts of energy and water. In Australia, projections indicate that these centers will triple the nation’s energy and water usage by 2030. At a time when the most effective approach to combat climate change involves rapid electrification through renewable energy and storage, permitting these energy-intensive facilities to burden the grid could hinder the shift toward net-zero emissions and inflate energy costs for consumers.
As fossil fuels continue to supply around half of our energy needs, the addition of significant greenhouse gas emissions remains a concern. Notably, Queensland has expressed its willingness to continue relying on fossil fuels for data centers, defying the federal government’s expectations. Any analysis of costs and benefits must factor in the cumulative impact of emissions produced by data centers.
Additionally, the issue of waste heat generated by data centers is considerable; the substantial energy input results in heat output. While this byproduct can be beneficial in colder regions, such as Finland, where it is utilized for residential heating, many parts of Australia are already experiencing increased occurrences of extreme heat due to climate change.
Regarding economic growth and job creation, while the data center boom has indeed spurred business investment in the past year, most equipment needed for these centers is imported. Consequently, the direct impact of this investment on the overall economy is minimal. Beyond the construction phase, data centers generate significantly fewer jobs compared to other sectors, such as manufacturing.
When Australian politicians or industry advocates discuss the advantages of data centers, they are primarily referring to the potential benefits derived from the AI technologies these facilities support, particularly the productivity improvements anticipated from AI, regardless of their timing or magnitude. In a recent address to the Australian Business Economists, Assistant Minister for Science, Technology, and the Digital Economy Andrew Charlton remarked that Australia stands at a pivotal moment. The country can either remain a “technology taker” enjoying some productivity gains or strive to become “a world-class adopter, creator, and exporter of AI technology.” Given Australia’s historical challenges in commercializing innovations and retaining profits domestically, achieving the latter will undoubtedly be more complex.
Charlton further emphasized the need for the government to ensure that technology serves the Australian populace, rather than the reverse. A review of the current landscape of data centers and AI, along with their associated costs, suggests that this goal has yet to be realized.
Nicki Hutley is an economist specializing in consulting.
