Australia is hosting an AI Boom: but who is getting paid?

Australia is fast becoming a serious destination for artificial intelligence infrastructure.

Billions of dollars are being directed toward data centres, compute clusters and energy-intensive AI systems. Proposals such as the reported $7 billion OpenAI–NEXTDC campus in Western Sydney highlight the scale of investment now in play. At first glance, this looks like a clear economic win.

But a more important question sits beneath it: who is actually capturing the value?

If you examine how AI infrastructure works, the answer is not Australia, at least not in the way many assume. Large data centres create visible economic activity. There is construction, engineering, local contractors and supply chains. During operations, there are utility payments, maintenance services and a relatively small number of permanent jobs. These benefits are real. But they are not where the long-term value lies. The real economic returns in artificial intelligence are generated by intellectual property, software platforms, enterprise contracts, and recurring revenues built on top of them. These assets drive profitability and equity value. And they are overwhelmingly owned by global technology companies.

So while Australia hosts the infrastructure, the core economic upside is likely to flow offshore. Australia has encountered this problem before. We are a resource-rich country, but the extent to which those resources have been converted into long-term national wealth has always been debated.

Compare Australia with Norway.

Norway did not simply extract oil and gas. It built institutions to capture the value and invested it through a sovereign wealth fund now worth more than US$1 trillion. Australia took a different approach. Our resource sector is highly successful, but the mechanisms for capturing and compounding national wealth have been more limited. The risk now is that we repeat that pattern in a new sector.

Artificial intelligence infrastructure is not just another technology cycle. It is emerging as a strategic asset class, combining physical infrastructure, energy systems and long-term capital. Countries that get this right will not just host AI systems. They will participate in the value those systems generate. Countries that get it wrong will provide the land, power and regulatory environment while the economic upside accrues elsewhere.

Australia is now at that point. The answer is not to deter investment. Australia should position itself as one of the best places in the world to build AI infrastructure. But it needs a clearer framework to ensure it shares in the benefits:

• first, develop a market-based system for licensing training data, so Australian content is not used without compensation;

• second, invest in sovereign compute capability, ensuring domestic researchers and industries have secure access to AI infrastructure;

• third, enable Australian participation in the assets themselves, including through superannuation and institutional capital.

None of these measures are anti-investment. They are about structuring it intelligently and result in Australia attracting AI infrastructure investment.

The real question is whether it becomes a host or a participant?

If policy settings remain unchanged, the likely outcome is familiar: infrastructure built locally, value realised globally. But with the right framework, Australia can do more than facilitate the AI boom.It can share in the wealth it creates.