In this episode of PING, APNIC Chief Scientist Geoff Huston reflects on the economic consequences of the surge in artificial intelligence (AI) investment and the associated data centre build-out.
Planned AI and data centre investment is measured in the hundreds of billions, and potentially trillions, of dollars globally. Geoff argues that this level of capital allocation is becoming economically significant at a national scale.
Technology consumers are already familiar with the effects. Prices for key components such as RAM, storage devices, and graphics processing units (GPUs) have risen sharply, while supply remains constrained. However, AI investment’s impact reaches far beyond the IT sector. When speculative activity accounts for such a large share of economic activity, any slowdown, correction, or collapse can divert capital away from productive investment elsewhere in the economy.
Much of the current AI investment is based on expectations of future returns. As a result, some investments carry significant risk. The expected returns may never materialize, but the debt used to fund data centre construction will remain. If projects fail, investors must absorb the losses. Debt may be recovered through asset sales at reduced prices, or losses may be distributed among those who financed the projects.
For most people, a concern is that these investment decisions are largely made on their behalf. While risk tolerance varies between individuals, banks, investment firms, and institutional investors make many of the decisions that determine where capital is allocated.
These institutions may seem distant, but their influence is substantial. Australia’s superannuation sector manages more than AUD 4 trillion, making it the world’s fourth-largest pension investment pool despite serving a population of only 27 million people. As a result, Australians may have greater exposure to this risk than many realize.
The same issue applies to pension funds in Canada, the United Kingdom, the European Union, and across Asia. It is particularly relevant in the United States, where AI companies have become, or are close to becoming, standard components of diversified investment portfolios.
Building and operating a modern AI-focused data centre requires major investment in construction, power infrastructure, cooling systems, and supporting facilities. The profitability of these projects allows operators to compete aggressively for labour and materials, often paying more than other sectors. This is contributing to supply chain constraints across civil engineering and is affecting unrelated construction activities, including housing.
The effects also extend beyond construction. Growing demand from data centres is increasing pressure on electricity networks and contributing to higher power costs for households and businesses. The location of these facilities can also affect surrounding communities, influencing their amenity and long-term viability as places to live and work.
As Geoff notes, these facilities have become so large that, in the United States, two competing projects of similar scale often cannot be supported within the same state. They must be located separately to avoid overwhelming local utility infrastructure.
This situation was not the result of deliberate global economic planning. Instead, it has emerged rapidly and continues to gather momentum. The challenge is global. As investment grows, concerns about risk exposure are increasing, particularly as weaknesses in current business models become more apparent.
The question of who ultimately wins or loses extends beyond individual AI companies or data centre operators. If the current investment boom proves unsustainable, the resulting economic shocks could affect investors, consumers, and communities far beyond the AI sector.
Read more about the economics of AI on the web:
- The Economics of AI (Geoff Huson, APNIC Blog)
Subscribe and share your story
You can stream and subscribe to PING via the following channels:
If you’re interested in sharing your insights or research, please get in touch — we’re always looking for great stories from the community. Please let us know what you think of the podcast and the APNIC Blog so we can keep improving.
The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.