Artificial-intelligence computing infrastructure is more critical to Australia’s strategic future than developing sovereign AI models.

The country should indeed aim at becoming a dependable computing hub for its allies.

Markets forecast that AI systems could match or exceed human performance in most cognitive tasks within two years. Even if this timeline proves optimistic, a transformative impact on the global economy is certain, akin to the agricultural or industrial revolutions. Treasurer Jim Chalmers acknowledged this, saying, “the future of our economy will come from … the artificial intelligence revolution.”

Australia must not misjudge this moment. AI is on track to result in a significant wealth transfer from AI takers to AI makers. Without a foothold in the AI value chain, Australia risks being relegated to the economic periphery.

Currently, Australia plays only a minor role in this ecosystem. Australians are predominantly AI consumers, relying on overseas jurisdictions for foundational models and computing infrastructure while bleeding data and talent. In response, some advocate for developing sovereign AI models. But this path risks a costly arms race. Training frontier models costs hundreds of millions of dollars today and may soon cost hundreds of billions. There’s no prize for second place in the model race.

Instead, Australia should heed the gold rush maxim: there will always be demand for shovels. Our country’s political stability, robust rule of law and strong alliances provide a solid foundation. Combined with renewable energy potential, Australia is uniquely positioned to become a trusted global hub for AI computing infrastructure, particularly sustainable data centres optimised for next-generation AI workloads.

Demand for computing resources is booming. The hundreds of billions of dollars a year on model training are going to get spent somewhere. AI model training increasingly relies on large-scale infrastructure, and global players are scrambling to secure access to hardware and energy. The US recently partnered with Gulf states such as Saudi Arabia and the UAE to build computing facilities. Gulf states may have abundant energy and lighter regulatory constraints, but building critical AI computing resources in non-aligned states risks undermining democratic security. As Jordan Schneider of ChinaTalk notes: ‘If [computing power] really is as important as we believe, why are we letting these countries and companies we deeply distrust get access to it?’

Computing infrastructure also gives as an avenue to shape global norms and rules around AI. Speaking to Australia’s potential as a computing-resources hub for companies like OpenAI, Western Sydney University’ James Arvanitakis stressed that AI and computing infrastructure must integrate ethical and societal values, ensuring technological progress aligns with broader public interests.

Hosting critical AI infrastructure gives Australia at least some strategic leverage to influence AI safety and global governance norms—through conditions of use, trusted partnerships and participation in international standard-setting.

There are other arguments for Australia being a global hub for computing resources. Old concerns about latency no longer apply. The fractional-second delay in global communication that held back Australia’s ability to lead in cloud storage is irrelevant for most AI workloads. Additionally, nearly 70 percent of people globally are concerned about the environmental footprint of AI. With strong social licence for clean energy and plans to increase its contribution to generation by 33,000 gigawatt-hours each year, Australia is well placed to offer secure, sustainable and scalable computing resources to the world.

Australia’s potential as a global computing-resources hub is constrained by slow planning approvals, limited energy grid capacity and insufficient investment in off-grid renewables. These issues are compounded by stronger regional competition and a lack of clear national strategy or incentives to attract major AI infrastructure projects. However, targeted government investment in energy infrastructure, streamlined regulatory processes, strategic policy support and local innovation could unlock Australia’s competitive advantages.

Strategically, this matters. Taiwan’s semiconductor industry has been called its silicon shield, since global dependency on it deters conflict. Australia could build a silicon spine—a decentralised, renewables-powered computing backbone that supports allied AI development. This would give Australia the economic relevance and strategic leverage it needs in the AI era.

Computing resources alone won’t solve every challenge Australia faces. Complementary investments in local models, data governance, safety and local innovation remain appropriate. But securing a place in the global computing-resources supply chain gives Australia a spine to graft other initiatives onto. Investing in infrastructure is economically safer than gambling on models and gives us a way to move forward commercially, geopolitically and environmentally.