- Posted on 23 May 2025
This article appeared in East Asia Forum on May 23 2025
Artificial intelligence (AI) is reshaping the world, in fields as diverse as healthcare and climate modelling to advanced manufacturing and defence. But Australia’s AI transformation is becoming increasingly geopolitically charged under the unfolding US–China tech rivalry. Australia faces the challenge of harnessing AI’s potential through international partnerships, without ceding its technological sovereignty.
Research partnerships between the United States and China remain the largest globally, with the two nations leading innovation and development in artificial intelligence. Since 2010, US–China AI collaborations have grown roughly fourfold — more than twice as much as the next leading country–pair.
A US–China ‘dual-core structure’ has crystallised since 2016, hardened by restrictions on talent mobility, curtailed research exchanges and widening ideological divides amid an escalating tech rivalry. Under the Biden administration [1], the United States pressed allies to decouple from Chinese technology on national-security grounds, adopting a ‘small yard, high fence’ strategy that tightens export controls on AI and semiconductors.
Trump 2.0 has pushed this further [2], mandating vetting of bilateral research projects, disbanding independent science advisory panels, restricting visas for Chinese scholars and curbing university partnerships. Australian researchers reliant on US funding now face growing pressure to disclose ties with Chinese institutions and certify alignment with US policy priorities — ranging from national security protocols to diversity, equity and inclusion mandates.
These new measures are fracturing long-standing research networks, accelerating the decoupling of the dual-core AI collaboration system and forcing Australian scholars to ‘choose sides’ [3] — undermining academic freedom, stifling innovation and severing valuable research partnerships with scholars across both major AI powers.
As a smaller player in the global AI ecosystem, Australia’s close ties with both the US and China are crucial to its research capacity [4]. Australia’s long-standing alliance with the United States embeds it within Western research networks [5]. But strong economic ties and expanding exchanges in research and education have deepened collaboration with China. In fact, research links between Australia and China have outpaced Australia–US partnerships since 2005 and contribute disproportionately to ‘high-impact work’, underscoring the importance of China to Australia’s AI research [6] [7].
Australia’s AI sector reveals a stark contrast between research excellence and economic impact. In the 2024 Stanford Global AI Power Rankings, Australia ranked 16th in ‘AI vibrancy’ and 7th in ‘responsible AI’, but only 30th in ‘economic competitiveness’ among 36 countries [8]. Structural barriers continue to limit the nation’s ability to fully capitalise on its AI potential, including opportunities for collaboration with China.
Infrastructure remains a key constraint. Without large-scale domestic GPU clusters, Australian researchers depend on foreign cloud providers — raising costs and exposing sensitive data [9]. The National Computational Infrastructure Australian Intelligence platform offers some relief, but fragmented funding still prioritises short-term pilots over foundational research [10]. Open-source, low-cost models such as China’s DeepSeek could democratise AI access, yet engagement remains risky — security concerns have led to bans on platforms including DeepSeek and TikTok in sensitive research contexts [11] [12] [13] [14][15].
Data governance adds another layer of constraints. Strict OAIC privacy rules protect individual rights but limit cross-sector data sharing, while Australia’s small, homogenous population restricts dataset diversity in areas such as specialised medicine [16] [17]. Cybersecurity concerns over sharing sensitive data with partners such as China further complicate international collaboration [18].
Human capital remains a critical bottleneck. Australia must quintuple its AI workforce by 2030 to meet demand, yet competitive salaries abroad and restrictive visa policies fuel a brain drain to both US and Chinese tech hubs [19]. Recent restrictions on granting visas to foreign students and scholars, along with tighter oversight of defence-related research, risk inadvertently undermining the AI talent pipeline [20] [21]. This will weaken the development of future domestic AI hubs that sustain international collaboration networks.
Under renewed pressure from Trump, visa restrictions could further deter AI scientists with Chinese ties from collaborating with or studying and working in Australia — narrowing the talent pipeline as global competition intensifies.
Policy fragmentation and underinvestment further weaken Australia’s position. The Australian Research Council has reduced STEM grants involving Chinese partners, while the 2025 federal budget provided no new targeted funding for AI or digital infrastructure [22] [23]. Continued dependence on multinational cloud providers risks driving up costs, creating regulatory vulnerabilities.
Decoupling from China may seem like an easy solution. But Australia’s AI strength depends on maintaining collaboration with both Washington and Beijing. Leaning too close to Washington leaves Australia at risk of becoming further embedded in US-controlled data infrastructures, while leaning too close to Beijing may risk alienating the nation’s closest security partner. But sitting on the sidelines guarantees only one outcome — Australia falling behind the global technology frontier and becoming a passive adopter of AI innovations and applications developed elsewhere [24].
Australia must pursue a dynamic balancing act. AI research should be segmented into sensitive streams. Areas such as defence, critical infrastructure and dual-use projects should be subject to national security vetting protocols, while non-sensitive fields such as agriculture, climate science and health should remain open for collaboration with China under transparent frameworks covering data use, intellectual property rights and exit clauses. Australia should also expand the National AI Capability Plan into a federated network of home-grown cloud infrastructure, supported by immigration policies designed to attract and retain top talent and encourage research partnerships [25].
Finally, Australia should leverage its reputation to help shape global AI governance. By co-designing open-science consortia and ethical-AI frameworks, Canberra can promote shared standards on transparency, data ethics and algorithmic accountability — protecting its researchers against bloc-based decoupling. In doing so, Australia can secure its place at the forefront of global AI development by simultaneously harnessing US strengths in foundational research and China’s scale, cost efficiency and diverse datasets.
[1] https://cepa.org/article/bidens-final-global-chip-controls-target-china-and-allies/
[2] https://theconversation.com/trump-is-surveying-australian-academics-about-gender-diversity-and-china-what-does-this-mean-for-unis-and-their-research-252282
[3] https://www.crikey.com.au/2025/03/21/australian-universities-donald-trump-china-usa-academic-freedom-antisemitism/
[4] https://www.industry.gov.au/sites/default/files/2024-07/AI%20ecosystem%20report%20Mar%202023%20Catalysing%20an%20AI%20Industry%20PDF.pdf
[5] https://www.csiro.au/en/research/technology-space/ai/nsf-ai-research
[6] https://eastasiaforum.org/wp-content/uploads/2025/05/Screenshot-2025-05-06-at-11.39.17%E2%80%AFpm.png
[7] https://eastasiaforum.org/wp-content/uploads/2025/05/Screenshot-2025-05-06-at-11.39.35%E2%80%AFpm.png
[8] https://hai.stanford.edu/news/global-ai-power-rankings-stanford-hai-tool-ranks-36-countries-in-ai
[9] https://www.lowyinstitute.org/the-interpreter/public-goods-case-australia-s-digital-sovereignty
[10] https://nci.org.au/news-events/news/nci-unveils-australian-intelligence-ai-a-revolution-data-analytics-and-ai-powered
[11] https://eastasiaforum.org/2025/04/16/chinas-deepseek-is-not-the-end-of-silicon-valley/
[12] https://www.nytimes.com/2025/02/10/world/asia/china-deepseek-education.html
[13] https://www.scimex.org/newsfeed/expert-reaction-australian-federal-government-bans-chinese-ai-deepseek-on-devices
[14] https://www.bbc.com/news/articles/c8d95v0nr1yo
[15] https://ministers.ag.gov.au/media-centre/tiktok-ban-government-devices-04-04-2023
[16] https://www.oaic.gov.au/privacy/australian-privacy-principles/read-the-australian-privacy-principles
[17] https://www.mja.com.au/journal/2025/222/1/inclusivity-australian-population-surveys-missed-opportunities-understand-health
[18] https://www.aspistrategist.org.au/as-chinas-ai-industry-grows-australia-must-support-its-own/
[19] https://techcouncil.com.au/newsroom/ai-to-create-200000-jobs-in-australia-by-2030/
[20] https://www.bbc.com/news/articles/cd734wed3y9o
[21] https://go8.edu.au/media-release-enabling-critical-defence-research-collaboration-will-underpin-national-security
[22] https://www.timeshighereducation.com/news/australian-research-collaboration-china-freefall?utm_source=chatgpt.com
[23] https://www.lowyinstitute.org/the-interpreter/public-goods-case-australia-s-digital-sovereignty
[24] https://eastasiaforum.org/2025/04/14/whoever-controls-the-ai-stack-controls-the-future/
[25] https://www.minister.industry.gov.au/ministers/husic/media-releases/australian-first-ai-plan-boost-capability
Written by

Marina Yue Zhang
Yanhui Wei, Shengpeng Wang, Zuanxu Chen