For most of the past two years, Australia’s public conversation about artificial intelligence has circled the same handful of anxieties: jobs, misinformation, copyright and the question of whether the technology can be trusted at all. A new intervention from within the Labor caucus wants to shift that frame entirely, arguing that the more urgent risk is not what AI does to Australia but what happens to the country if it hangs back while allies and rivals race ahead.
In an opinion piece published by The Nightly, Victorian Labor senator Raff Ciccone makes the case that Australia must “jump into the AI capability arms race” rather than treat the technology as a productivity add-on to be regulated at leisure. The framing is deliberately martial, and it lands at a moment when governments around the world are folding AI into their defence, industrial and economic security strategies rather than leaving it to the market.
From productivity story to strategic contest
The core of Ciccone’s argument is a reframing. Much of the domestic debate has treated AI as a tool that businesses adopt to work faster or cheaper, with the policy job being to manage the downside. He wants it understood instead as a strategic capability, in the same category as energy security, advanced manufacturing and defence industry, where falling behind carries a national cost that no individual firm will price in on its own.
That distinction matters because it changes who is responsible. If AI is just another software category, government’s role is light-touch: set some guardrails and let companies buy what they need. If it is a capability that underpins economic and military strength, then the state has a direct interest in whether Australia can build, train and run advanced models on home soil, and in whether the compute, data and talent to do so exist within reach. Ciccone’s column pushes firmly towards the second view, and he is not alone in the parliamentary Labor party in doing so.
You can read the full argument in his piece for The Nightly, which sits alongside a growing body of commentary pressing Canberra to move faster.
Why the “arms race” language is doing work
The choice of words is not accidental. Calling something an arms race implies that standing still is itself a decision with consequences, and that the pace is being set by others. The United States and China are pouring hundreds of billions into frontier models and the data centres that feed them, while allies such as the United Kingdom, France, Japan and the United Arab Emirates have all announced sovereign AI programs aimed at keeping critical capability inside their borders. Against that backdrop, a middle power that treats AI purely as an import risks becoming a permanent customer of technology it cannot itself produce or fully understand.
For a senator, the rhetorical move also has a domestic audience. Framing AI as a race helps unlock the kind of spending and coordination that a productivity story rarely justifies. It is far easier to argue for sovereign compute, defence AI projects and industrial support when the alternative is described as falling behind competitors than when it is described as helping companies write emails faster.
The other side of the argument
Not everyone accepts the premise. Critics of arms-race framing, including many researchers who work on AI safety, warn that treating capability as a contest can crowd out the harder questions about reliability, bias, transparency and control. Their concern is that if every dollar and every headline goes to building bigger and faster systems, the unglamorous work of making them safe and accountable gets deprioritised, and that a race dynamic can push governments to relax the very standards that protect citizens.
There is also a hard-nosed economic objection. Sceptics point out that Australia cannot realistically out-spend the United States or China on frontier models, and that pouring public money into a capability race the country cannot win at the top end may be less sensible than picking narrow niches where local expertise, data and industry actually give an edge. That view has been argued forcefully by figures such as CSIRO’s Doug Hilton, who has made the case that Australia’s advantage lies in specialised models built around its own strengths rather than in trying to match the hyperscalers head to head. Between the “jump in and race” school and the “choose your niches” school sits most of the practical policy debate.
The Australian stakes
What makes Ciccone’s intervention worth watching is that it comes from inside government rather than from a think tank or an industry lobby. The Albanese government has been assembling the scaffolding of an AI agenda, from an emerging national capability plan to fresh interest in defence applications, and internal advocates help shape how ambitious that agenda becomes. A senator publicly demanding that Australia treat AI as a strategic priority adds political weight to the argument for sovereign infrastructure, local model development and defence-focused programs.
The stakes are concrete. Australia is already a significant destination for AI data centre investment, with billions committed to new compute capacity, yet much of the intellectual property, the frontier models and the chips that make those facilities useful are controlled offshore. The country trains strong AI researchers but loses many of them to better-funded labs abroad. And in defence, where allies are moving quickly to embed autonomy and machine intelligence into everything from surveillance to logistics, a lag in capability translates directly into a lag in national security. Ciccone’s column is, in effect, an argument that these gaps will not close on their own and that market forces alone will leave Australia dependent.
There is a counter-risk that deserves equal billing. Pouring money into a capability race without the workforce, the energy supply or the governance to support it can produce expensive infrastructure that sits underused, and the surge of AI spending globally has already prompted warnings about a bubble. The challenge for any Australian strategy is to be ambitious enough to matter while disciplined enough to avoid chasing prestige projects that deliver little.
What’s next
The immediate question is whether this framing gains traction inside the government’s forthcoming AI plans. If the “capability” language filters into official strategy, expect it to show up as concrete commitments: sovereign compute funding, tighter links between AI research and defence industry, incentives to keep talent onshore and a clearer position on where Australia chooses to compete rather than merely consume. If it does not, Ciccone’s piece will stand as one more prod in a debate that is slowly shifting from caution towards ambition.
Either way, the terms of the conversation are moving. For most of the AI era Australia has debated how to govern a technology it largely buys. The senator’s contribution is a bet that the more important question is whether the country can build one, and that the window to decide is narrowing.
Sources: The Nightly.
