Executive summary
The global artificial-intelligence race is increasingly becoming a competition between technological ecosystems. In August 2026, Reuters reported that Washington was preparing to tell 35 AI partners that participation in the US-led Pax Silica initiative may be incompatible with membership in China’s newly established World Artificial Intelligence Cooperation Organization (WAICO). Kazakhstan is currently the clearest test case: it joined Pax Silica in June and became a founding member of WAICO less than a month later (Reuters, 2026a).
Introduction
Washington has legitimate reasons to demand greater alignment in the AI race. AI supply chains involve advanced semiconductors, data centres, critical minerals and technologies with military applications. Trusted partners can therefore offer greater supply-chain security and reduce technology leakage. Yet making the wider partnership fully exclusive is likely to be counterproductive, particularly for countries such as Kazakhstan. Its economic geography makes cooperation with both China and the West valuable, while its long-standing foreign policy relies on avoiding excessive dependence on any one power. For Washington, the better strategy would be to reserve the most sensitive forms of technological cooperation for trusted partners without requiring countries to abandon every competing international framework.
Pax Silica is no longer simply an AI governance forum. The US initiative links AI development with semiconductor supply chains, critical minerals, data centres and investment. In June 2026, the AI Opportunity Partnership committed participating countries to encouraging investment in next-generation data centres and expanding access to AI compute, while members can participate in shared investment opportunities (US Department of State, 2026a; Reuters, 2026a).
Kazakhstan Case Analysis
For Kazakhstan, the benefits are tangible. It became the first Central Asian country to join Pax Silica on 25 June, presenting participation as an opportunity to develop AI infrastructure and high-tech manufacturing (Government of Kazakhstan, 2026). The relationship extends beyond AI. Washington and Astana also have a critical minerals cooperation framework covering exploration, processing and development, including a proposed $1.1 billion tungsten mining and processing project (IEA, 2026).
Yet on 16 July, Kazakhstan also joined 28 other countries in establishing China-backed WAICO. Beijing presents the organisation as a more open model of AI cooperation, particularly for developing economies, while promoting Chinese open-weight models and capacity-building programmes (Reuters, 2026b; Guey et al., 2026). President Xi Jinping simultaneously offered to deepen Chinese cooperation in artificial intelligence and the digital economy to support Kazakhstan’s digital transformation (Chinese Ministry of Foreign Affairs, 2026).
Washington now sees membership in both organisations as potentially incompatible. China responded on 19 August, arguing that countries should retain the right to choose AI partnerships in line with their own development needs (Reuters, 2026c).
Discussion
The US position has a clear strategic logic. Advanced AI does not depend only on software. It requires chips, computing infrastructure, energy and critical minerals. Pax Silica therefore seeks to build a trusted supply chain from mineral extraction to advanced models.
In such a system, trust matters. A country simultaneously embedded in a rival technological ecosystem could theoretically become a route through which restricted technology, technical knowledge or strategic inputs reach competitors. China’s dominance in several critical mineral processing markets and Washington’s attempts to diversify those supply chains further strengthen the case for closer coordination among partners (Reuters, 2026a).
There is also a potential benefit for participating countries. Alignment with the US ecosystem could offer greater access to Western capital, advanced computing infrastructure and investment projects. For economies attempting to develop AI industries rather than simply consume foreign models, these opportunities are significant.
But this only makes exclusivity attractive if those benefits outweigh the costs countries must sacrifice.
Kazakhstan demonstrates the problem with applying the same loyalty test to emerging economies as to longstanding US allies.
Its foreign policy has historically been deliberately multi-vector, balancing relationships with China, Russia, the United States and Europe rather than allowing dependence on a single external power (Nyshanbayev et al., 2025). This strategy reflects economic realities rather than simple diplomatic ambiguity.
China is deeply embedded in Kazakhstan’s economy. In 2025, it accounted for around 19.2% of Kazakh exports and 29.2% of imports. By comparison, the United States accounted for only around 3.3% of Kazakhstan’s imports (Astana Times, 2026). Chinese involvement also extends to Kazakhstan’s digital infrastructure: the government has signed cooperation agreements with Huawei covering cloud technology, intelligent infrastructure, and telecommunications development (Qazinform, 2026).
Choosing the American ecosystem, therefore, carries an opportunity cost that countries such as Japan or Australia do not face to the same extent. Kazakhstan could gain access to US technology and investment while losing access to Chinese finance, infrastructure, affordable AI models, or commercial partnerships.
Conversely, forcing Astana to choose could produce exactly the outcome Washington wants to avoid. If US access comes with broad political conditions while China offers technology without demanding technological exclusivity, Beijing may become the easier partner. China is already explicitly presenting its AI diplomacy as an alternative oriented towards developing countries and technological accessibility (Reuters, 2026b).
The wider risk is that a coalition intended to reduce Chinese technological influence becomes so restrictive that strategically useful partners decide not to join.
Conclusion
This does not mean Washington should treat all dual participation as harmless. Some AI technologies genuinely create national security risks. Access to advanced chips, sensitive models, cybersecurity capabilities, or classified supply-chain information can reasonably depend on strict safeguards.
But technological security does not require complete geopolitical exclusivity.
Pax Silica could instead distinguish between levels of cooperation. Highly sensitive technologies could remain available only to countries that meet strict security, export-control and end-use requirements. Less sensitive cooperation—research, AI skills, data-centre investment, governance discussions or basic infrastructure—could remain compatible with participation in broader organisations such as WAICO.
Such an approach would make membership conditional on behaviour rather than simply on institutional affiliation. It would also allow Washington to compete by making its technological ecosystem more attractive rather than requiring partners to demonstrate loyalty by withdrawing from alternatives.
For Kazakhstan, continued diversification remains rational unless either side offers benefits large enough to compensate for the loss of the other relationship. For the United States, forcing that choice prematurely risks turning strategic partners into reluctant allies—or driving them towards China.
The AI race may increasingly divide the world technologically. Washington does not need to accelerate that division itself.
Bibliography
Astana Times (2026) ‘Kazakhstan’s Foreign Trade in 2025: Partners and Trends’, 21 April. Available at:
https://astanatimes.com/2026/04/kazakhstans-foreign-trade-in-2025-partners-and-trends/
Chinese Ministry of Foreign Affairs (2026) ‘President Xi Jinping Meets with President of Kazakhstan Kassym-Jomart Tokayev’, 16 July. Available at:
https://www.mfa.gov.cn/eng/xw/zyxw/202607/t20260716_11984340.html
Government of Kazakhstan (2026) ‘Kazakhstan Joins the Pax Silica Initiative to Develop AI Infrastructure and High-Tech Manufacturing’, 25 June. Available at:
https://www.gov.kz/memleket/entities/maidd/press/news/details/1248924?lang=en
Guey, W., Bougault, P., Zhang, W., de Moura, V.D. and Gomes, J.O. (2026) ‘World Artificial Intelligence Cooperation Organization (WAICO): Mapping an Emerging Institution in the Global AI Governance Regime Complex’, arXiv, 22 June. Available at:
https://arxiv.org/abs/2606.23860
International Energy Agency (IEA) (2026) ‘US–Kazakhstan Memorandum of Understanding on Critical Minerals Cooperation’. Available at:
https://www.iea.org/policies/29736-us-kazakhstan-memorandum-of-understanding-on-critical-minerals-cooperation
Nyshanbayev, N., Bekov, K., Baizakova, Z. and Amangeldy, A. (2025) ‘The Republic of Kazakhstan’s Multi-Vector Foreign Policy’, New Perspectives, 33(1), pp. 43–63. Available at:
https://journals.sagepub.com/doi/full/10.1177/2336825X241308432
Qazinform (2026) ‘Kazakhstan Signs Cooperation Memorandum with Huawei at Mobile World Congress in Barcelona’, 6 March. Available at:
https://qazinform.com/news/kazakhstan-signs-cooperation-memorandum-with-huawei-at-mobile-world-congress-in-barcelona-77a241/
Reuters (2026a) ‘US to Tell Partners They Must Pick Sides in AI Race with China’, 14 August. Available at:
https://www.reuters.com/world/china/us-tell-partners-they-must-pick-sides-ai-race-with-china-2026-08-14/
Reuters (2026b) ‘Xi Pitches China as Leader of New Global AI Order, Challenging US Dominance’, 17 July. Available at:
https://www.reuters.com/world/asia-pacific/chinas-xi-promotes-chinas-commitment-ai-access-speech-shanghai-conference-2026-07-17/
Reuters (2026c) ‘China Urges Respect for Digital Sovereignty in AI Race’, 19 August. Available at:
https://www.reuters.com/world/china/china-urges-respect-digital-sovereignty-ai-race-2026-08-19/
US Department of State (2026a) ‘Joint Statement on AI Opportunity Partnership’, 25 June. Available at:
https://www.state.gov/releases/under-secretary-for-economic-affairs/2026/06/joint-statement-on-ai-opportunity-partnership
US Department of State (2026b) ‘Outcomes of the Second Pax Silica Summit’, 26 June. Available at:
https://www.state.gov/releases/office-of-the-spokesperson/2026/06/outcomes-of-the-second-pax-silica-summit