ASIA
13 September 2026
Kevin Rudd on Australia's security and AI
This article is based on Kevin Rudd's interview with the ABC. It is similar to that on his speech to Australia's Press, but contains some more detail on AI.
Australia needs to urgently prepare for three national security challenges – China, AI and climate change. But, regrettably, Australia has drifted to become a complacent country, when, in fact, according to Kevin Rudd, there is a one in three chance of China invading Taiwan.
AI is a major national security issue bearing down on Australia. AI is a two- horse race right now between the US and China.
Who's winning? It depends on the matrix for analysis. The AI stack is a complex equation. It's about chips, computing power and speed. It's about AI infrastructure, and frontier models. It’s also about diffusion across the economy, which can lift productivity across the various sectors.
If you look across the stack on chips, the Americans remain ahead largely because of TSMC (the Taiwanese chip manufacturer). Intel and Qualcomm chips are also advanced. Nevertheless, China is closing the gap.
American frontier models, particularly those by Anthropic are more developed, enhanced and capable than those of the Chinese. But only by a factor of 3 to 5%. And this race changes almost by the month.
When it comes to the diffusion rate of AI into the real economy, the Chinese are ahead by about two to one in terms of the number of firms taking up AI in their mainstream businesses, producing what Xi Jinping hopes to be the ultimate productivity yield.
Why does this race matter for Australia? Will we at some point have to choose one side or the other when it comes to AI or have we already made that choice?
The Australian AI industry is firmly linked with the US. There are a range of not just geopolitical reasons for that, but also our ability to access the world's fastest and best chips is predicated on our status as a reliable “five eyes” partner.
But there are also Australian businesses using cheaper Chinese AI products. They have been diffused more generally.
But the key is that if we don't develop our own AI industry, our ability to generate effective and rolling AI defenses (cyber defenses) against each new frontier model which emerges will be compromised. We will be totally dependent on others providing those defenses to us.
The challenge is to build a capacity to generate not just wealth, but also the technology base to generate the rolling set of defenses we're going to need.
We have clever Australians, but do we have the money that you would need for frontier models here? Already in the last three years we've secured some $30-35 billion worth of direct investment across the AI stack from companies like Amazon, Microsoft, Google and Anthropic.
Asia Society analysis suggests that there are tens of billions of further investment waiting in the queue to come to this country as well. This will enable us not just to build our own national compute capabilities. Once the compute is developed and you have the workforce training skills, you can then move rapidly into the development of frontier models as well. You can’t do the latter without the former.
What about the community reaction to data centres? Rudd believes that once the government has framed its final AI policy, based on the Prime Minister's recent speech, the answer will be positive. But it will be necessary to build safeguards
If we can be an attractive place for the data centres, that will give us leverage to be able to access the models that we're going to need for our cyber security. And leverage not just to access them, but to develop them, co-develop them, then develop them. We must become an AI power in our own right for national security reasons. Australia must have this set of capabilities ourselves, sovereign capabilities, not least because we need to defend ourselves.
AI is a major national security issue bearing down on Australia. AI is a two- horse race right now between the US and China.
Who's winning? It depends on the matrix for analysis. The AI stack is a complex equation. It's about chips, computing power and speed. It's about AI infrastructure, and frontier models. It’s also about diffusion across the economy, which can lift productivity across the various sectors.
If you look across the stack on chips, the Americans remain ahead largely because of TSMC (the Taiwanese chip manufacturer). Intel and Qualcomm chips are also advanced. Nevertheless, China is closing the gap.
American frontier models, particularly those by Anthropic are more developed, enhanced and capable than those of the Chinese. But only by a factor of 3 to 5%. And this race changes almost by the month.
When it comes to the diffusion rate of AI into the real economy, the Chinese are ahead by about two to one in terms of the number of firms taking up AI in their mainstream businesses, producing what Xi Jinping hopes to be the ultimate productivity yield.
Why does this race matter for Australia? Will we at some point have to choose one side or the other when it comes to AI or have we already made that choice?
The Australian AI industry is firmly linked with the US. There are a range of not just geopolitical reasons for that, but also our ability to access the world's fastest and best chips is predicated on our status as a reliable “five eyes” partner.
But there are also Australian businesses using cheaper Chinese AI products. They have been diffused more generally.
But the key is that if we don't develop our own AI industry, our ability to generate effective and rolling AI defenses (cyber defenses) against each new frontier model which emerges will be compromised. We will be totally dependent on others providing those defenses to us.
The challenge is to build a capacity to generate not just wealth, but also the technology base to generate the rolling set of defenses we're going to need.
We have clever Australians, but do we have the money that you would need for frontier models here? Already in the last three years we've secured some $30-35 billion worth of direct investment across the AI stack from companies like Amazon, Microsoft, Google and Anthropic.
Asia Society analysis suggests that there are tens of billions of further investment waiting in the queue to come to this country as well. This will enable us not just to build our own national compute capabilities. Once the compute is developed and you have the workforce training skills, you can then move rapidly into the development of frontier models as well. You can’t do the latter without the former.
What about the community reaction to data centres? Rudd believes that once the government has framed its final AI policy, based on the Prime Minister's recent speech, the answer will be positive. But it will be necessary to build safeguards
If we can be an attractive place for the data centres, that will give us leverage to be able to access the models that we're going to need for our cyber security. And leverage not just to access them, but to develop them, co-develop them, then develop them. We must become an AI power in our own right for national security reasons. Australia must have this set of capabilities ourselves, sovereign capabilities, not least because we need to defend ourselves.