AI is the most consequential technology we will ever create. The visit by Japan’s pro-industry Prime Minister Sanae Takaichi to Australia this week was a chance to bind our efforts with those of a natural friend in a field that will determine which countries carve out secure and prosperous futures and which limp into this new age.
Takaichi and Prime Minister Anthony Albanese signed an economic security agreement that pledged to ‘look for opportunities to further promote concrete cooperation’ on technologies including AI. It was a promising development. Now, the government should lean into this opportunity as far as Japan is willing to go. We have everything to gain by linking arms with a regional neighbour whose strategic outlook is the closest to Australia’s of any Asian nation and that has proven, when it puts its mind to it, that it can move technological and industrial mountains.
AI is dominated to a dangerous degree by two tech superpowers, the United States and China, which also happen to be the 21st century’s history-defining strategic rivals. Such concentration – particularly in frontier models such as Anthropic’s Claude, OpenAI’s GPT-5 and China’s DeepSeek – is not healthy for the rest of us.
China is looking to dominate AI and all the surrounding technologies from semiconductors to applications in a way that frees it of dependencies but makes others dependent on it. Beijing has shown it is willing to restrict access to its critical materials and technologies to bend another country to its will.
The US, meanwhile, will remain a vital partner to Australia and Japan – a fact both governments strenuously acknowledge – but in the wake of US President Donald Trump’s use of tariffs and the withdrawal of security cooperation as coercive measures to pull even close allies into line, we should not bank on the US always sharing its high-end technology. Washington’s priorities are likely to remain messy and unpredictable, as we’ve seen in its approval, at least in principle, of the sale of US semiconductor giant Nvidia’s advanced chips to China for training AI models.
Even if it has the best intentions, the US might not be able to supply its friends, given the rampant demand for cutting-edge semiconductors and the worryingly slender supply chains.
Australia is safer if it develops greater sovereign capability. Sovereignty doesn’t mean doing everything ourselves or vainly trying to match the US and China in areas such as frontier models. But with friends such as Japan, we can build a more diverse and secure AI sphere to ensure we have resilient access to core capabilities.
Japan is the right partner for several reasons. Takaichi has recognised that Japan’s strategic, demographic and productivity challenges demand lavish investments in advanced technologies, mobilising not just government money but also the backing of national giants such as Sony, NEC, Softbank and Honda. These companies have recently joined a consortium to develop a frontier model that can be used to interact with the real world – distinct from a chatbot – and that will mesh nicely with Japan’s advanced manufacturing and industrial supremacy, especially its dominion in robotics.
Showing a genuine sense of urgency, the Japanese government has set up an AI Strategic Headquarters reporting directly to the prime minister’s office. Fortune Business Insights has projected the Japanese AI market to grow from US$15.6 billion (A$21.66 billion) in 2025 to US$123 billion in 2032.
Japan is also investing heavily in semiconductors, setting up a state-backed flagship bet in Rapidus Corporation, with investment from Japanese conglomerates such as Toyota, Sony, Softbank and Fujitsu.
While very pro-development, Japan also champions ‘human-centred’ AI principles that resonate well with a lot of other countries and put Japan in a good spot to influence global governance towards the kind of democratic values that our two countries share.
Australia, for its part, brings critical minerals, strong potential for AI adoption in mining, financial services and healthcare, and its natural capacity to host data centres given its ample land, renewable energy and geography far from physical threats. Our two countries’ strengths are nicely complementary.
The cooperation could include sharing know-how through technology transfer; research collaboration; incentives for two-way investment; greater skills mobility through favourable visa arrangements for AI workers; research and development tax credits for joint AI projects; and purchase guarantees so that our sectors line up. For instance, Australia could guarantee to buy a portion of Japan’s semiconductor output and Japan could pledge in turn to buy a set amount of computing power from Australian data centres.
AUKUS Pillar Two – the initiative concerning not submarines but defence tech cooperation – is a useful model. Given Japan’s interest in joining Pillar Two, Australia–Japan AI cooperation could sit alongside AUKUS and gradually integrate with it where beneficial.
There’s precedent for this kind of strategic cooperation. Japan invested in Australian rare-earths firm Lynas to shore up supplies after Beijing in 2010 banned rare-earth exports to Japan in an effort to browbeat Tokyo over an unrelated territorial dispute between the two countries in the East China Sea. As Resources Minister Madeleine King told ASPI’s Stop the World podcast last month, ‘there is just a really high level of trust’ with Japan.
Australia’s decision to buy Japan’s Mogami-class frigates is another concrete example of the growing strategic closeness. AI cooperation wouldn’t be about autarky – that is, total self-sufficiency. We’ll still want the US’s unrivalled frontier models and chip design and for our ally to outcompete China. But embracing Japanese AI as firmly as possible would expand our strategic options, ensure access to capabilities indispensable to our security and economy, and strengthen our clout in our dealings with the superpowers.
This is a lightly edited version of an article originally published in The Sydney Morning Herald.
