Decoupling is well underway in critical technology research. A divide is emerging between China’s critical technology research ecosystem and that of the United States and its allies.

The implication is clear: to varying degrees, policies adopted since late last decade to guard against China’s exploitation of research in democratic countries are having an effect.

After decades of growth, US and Chinese technical research collaboration peaked in 2019, according to new research from the Critical Technology Tracker of the Australian Strategic Policy Institute (ASPI). Despite the fall since then, the US does remain China’s largest research partner, however.

The downward turn coincided with the launch of US policies that focused on national security risks in research collaboration, notably the Justice Department’s China Initiative. As collaboration with US researchers has diminished, Chinese researchers have strengthened research connections with some of China’s closest partners, including Pakistan, Saudi Arabia and Iran.

The move towards restricting collaboration has continued across three successive US administrations. The Biden administration renegotiated the US–China science agreement to narrow the scope of collaboration, removing provisions that encouraged deeper institutional ties beyond government channels. Congress is pushing to revive the China Initiative, citing concerns over ‘China’s malign ambitions to steal American research.’ Intended to investigate and prosecute China-linked cases of economic espionage and intellectual-property theft, the program’s initial implementation focused mainly on prosecuting academics for failing to properly disclose ties to Chinese institutions on federal research grant applications.

Both Beijing and Washington see science and technology as the bedrock of military strength, economic development and even cultural and political influence. In this new era of techno-nationalism, countries are vying for leadership in new and emerging technologies as a means of asserting global power.

The shift is quantified by data from ASPI’s Critical Tech Tracker—more than 7 million papers on technical advances across 74 critical technologies (a recently expanded data set that in the next few weeks will feed an updated edition of the tracker’s public version). Since 2005, China’s share of global research output has surged from 13 percent to nearly 40 percent. As China has moved from the periphery to the centre of global innovation, international collaboration with Chinese researchers has naturally grown.

Figure one shows that the US remains China’s single largest source of international research collaboration, but the turning point in 2019 is clear. The decline is even more severe after accounting for the overall growth in research volume over the two decades. The volume of publications for 2024 in the dataset is five times the figure for 2005, so growing research volume risks hiding a decrease in collaboration per unit of research.

To correct for this, we define collaboration intensity as the ratio of co-authored publications between the two countries to an average of the research outputs for both (see methodology for details). With this adjustment, the decline becomes even starker, as depicted in figure two.

Pakistan, for example, progressed from virtually non-existent collaboration in 2005 to become China’s seventh most intensive research partner in 2019, surpassing Japan. This rise was particularly driven by collaboration in nanomaterials research, a field where Pakistan now ranks eighth globally in high-impact research, ahead of Germany and the United Kingdom. Reflecting the growing momentum of their science and technology partnership, China and Pakistan signalled plans last month to deepen cooperation in quantum technologies. This adds to existing cooperation between the two countries in nuclear technologies, space exploration and artificial intelligence.

Other partners have similarly gained in position. If each European Union country is ranked separately, Saudi Arabia can be seen to have made an even more dramatic leap, climbing from 46th position in 2005 to eighth in 2024. On the same basis, Belarus rose from 41st to 20th while Iran moved from 35th to 25th. Russia, after an initial decline, began to recover in the late 2010s and climbed back to 22nd. This cooperation now extends beyond formal research to include technology transfer, venture capital and joint infrastructure development.

To visualise these fragmented shifts, figure three indexes collaboration intensity to 2005 levels, a benchmark year for open engagement between China and the West. The data reveals a fragmented global shift: while some countries have significantly strengthened their research ties with China, others have plateaued, and a growing number have begun to retreat.

The scale of this shift is significant. In 2016, nearly half of China’s international collaborations involved US scientists; that figure is now slightly more than one in four. When adjusted for intensity, US–China collaboration is drifting back to 2005 levels—an era when innovation was still centred in the Euro-Atlantic region.

This disengagement extends beyond the US, as other allies become cautious about risks to knowledge security.

Concerns over research security in Australia’s science and technology sector gained traction from 2018. That year, an ASPI report highlighted research collaboration between Australian universities and the Chinese military, often with seemingly civilian universities and supported by Australian government grants. Growing concerns prompted various initiatives between 2018 and 2024 to strengthen risk assessment frameworks. Concurrently, funding from Australia’s peak grants body, the Australian Research Council, for research projects involving China-based collaborators declined sharply, falling from a peak of A$90m in 2019 to A$33m in 2024.

China’s collaboration with other countries may be trending in a similar direction, although so far this retrenchment has manifested more as a plateau than a clear decline. In the UK, this levelling-off, beginning in 2021, coincided with the former Conservative government’s placement of research security ‘at the heart’ of its international research collaborations. As part of this shift, the government conducted a review of all research and innovation ties with China, mirroring a similar assessment done for Russia the previous year. In this context, funding for collaborative research projects with China from UK Research and Innovation, the country’s national grants agency, has just about evaporated. It peaked at £112m in 2016 but was a mere £400,000 in 2022, the agency says.

Despite maintaining a relatively conciliatory approach toward China as other countries have taken tougher stances, New Zealand has become increasingly vigilant against strategic research risks, leading to a plateau in scientific and technological engagement with China. This shift is evidenced by the 2023 renewal of the two countries’ bilateral science agreement, which removed previously listed subjects of nanotechnology and superconductors from its list of priority engagement areas. The remaining areas focus on only health, food and environmental sciences. In 2019, the New Zealand Security Intelligence Service also stepped up efforts to warn universities about the threats to knowledge security arising from international research collaborations.

Meanwhile, some countries, such as Italy, have steadily deepened their scientific cooperation with China since the early 2010s, as shown in figure five. Sino-Italian scientific collaboration, formerly minor, grew significantly after the 2009 European debt crisis, as Italy’s financially constrained research sector increasingly relied on China’s well-funded ecosystem. Researcher Lorenzo Mariani has described this as ‘an exemplary case study of China’s strategy to internationalise its scientific power’.

To underscore the geopolitical dimensions of this realignment, figure six aggregates all countries in our 20-year dataset into three distinct groups: China’s strategic partners, such as Russia, Pakistan, Belarus and Iran; its competitors, such as the US, Japan and the UK; and a third group comprising all other countries, termed ‘hedging countries’.

China’s research collaborations have historically leaned toward countries it now finds itself in strategic competition with, but since 2016 there has been a rapid shift toward geostrategic diversification. This recalibration appears driven by growing knowledge security concerns in the West and by China’s deliberate outreach to non-Western partners. While Western concerns have led to real policy changes, they also risk isolating allied nations from emerging innovation hubs.

The risk of bifurcation is not simply that democracies lose access to Chinese research; it is that they lose the ability to steer global technological development. Avoiding that outcome requires more than decoupling; it demands recoupling on new terms, built around trusted networks, shared standards and deliberate technological specialisation. Data-driven approaches such as those offered by ASPI’s Critical Tech Tracker can help rebuild trusted research networks, guide strategic investments and reinforce allied leadership in critical technologies that will define the 2030s and beyond.

Methodology:

The research in this report builds upon ASPI’s Critical Technology Tracker methodology. Bibliometric data was sourced from Clarivate’s Web of Science Core Collection, with research relating to critical technologies identified using ASPI-developed search terms.

In this special report, weighted collaboration is quantified using the bi-directional fractional counting method introduced by Loet Leydesdorff and Han Woo Park (2016), where weighted collaboration (c_p) between countries A and B for paper p is:

n_p is the total number of authors of paper p, n_p,A and n_p,B are the total number of authors from countries A and B respectively. This metric places a greater emphasis on more direct collaborations. For example, a paper co-authored by one researcher from the US and one from Australia represents a larger Australia–US collaboration than a paper that also includes three additional authors from Germany. The total weighted collaboration score C(A,B) sums across all papers each year.

Collaboration intensity I(A,B) is defined as the ratio of weighted collaboration to the geometric average of each collaborating country’s total research output:

R_A and R_B represent the total research output of countries A and B. These totals are calculated using the same fractional method as the Critical Tech Tracker. The geometric average is used—rather than the arithmetic average—to better reflect the balance of research outputs between the two countries. Unlike the arithmetic average, the geometric average reduces the influence of extreme differences. This makes it a fairer way to compare collaboration intensity, especially when one country produces significantly more research than the other. It helps ensures that the ratio reflects mutual engagement, rather than being dominated by the larger partner.