Quick Summary
- The report contrasts the US-led Pax Silica initiative with China’s World AI Cooperation Organization.
- Kazakhstan has joined both initiatives and is seeking investment in AI infrastructure and supply chains.
- Countries weigh advanced US chips and partnerships against lower-cost Chinese models and a greater role in setting standards.
Countries that want to expand artificial intelligence are facing a strategic problem: the United States and China increasingly present competing technology ecosystems, yet many governments want access to both. The Associated Press reported on October 25 that some countries are trying to benefit from the US-led Pax Silica initiative and China’s World AI Cooperation Organization rather than committing exclusively to either. The choices affect more than which model a ministry buys. They can shape access to chips, investment, infrastructure, technical partnerships and the standards that influence how AI systems are developed and deployed.
Kazakhstan is one example in the AP report. The country joined the US-led effort to strengthen AI supply chains and later became a founding member of China’s initiative. It has critical-mineral resources and relatively cheap electricity, both relevant to data-centre development. Earlier in 2026, its government signed an investment deal worth up to $10 billion involving Nvidia for a planned data-centre valley. The report also cites record Chinese foreign direct investment of $2.8 billion in Kazakhstan in 2025. These relationships create opportunities, but they also expose the country to pressure from both sides.
AI partnerships are becoming strategic infrastructure
The two initiatives are presented differently. US officials emphasise trusted partners, advanced chips and technology supply chains, while China promotes a more open and inclusive approach to AI cooperation. Those positions reflect the interests of the countries promoting them. The United States has a strong network of technology allies and leading AI infrastructure, while China is seeking partners for its models, hardware and standards. Each side’s messaging should be assessed alongside practical conditions attached to access, including export restrictions, data rules, procurement requirements and long-term support.
The issue is not limited to Central Asia. The AP report describes Indonesia seeking to benefit from both technology powers, in part because of its mineral resources and industrial ambitions. Singapore is considering joining China’s AI organisation while attracting US AI investment, including an Asia office from Anthropic. Saudi Arabia is using a Chinese open model in one platform while also working with Nvidia on computing infrastructure. France is trying to build an independent position through its AI industry, while Japan remains closely aligned with the United States on strategic questions but may still consider more affordable Chinese models for lower-risk applications.
For governments, the choice of AI stack can become a question of sovereignty. A country may want to host its own data, keep sensitive workloads within its borders and develop local expertise, but it may not have enough capital or electricity to build frontier models at the same scale as the US or China. Open-weight models can reduce some dependence on a single supplier, though they still require computing resources, maintenance and technical capability. Proprietary systems may provide strong performance or support, but access can depend on commercial terms and political decisions outside the buyer’s control.
Standards are another part of the competition. Countries that join an initiative may gain a voice in discussions about safety, interoperability and responsible deployment, but they may also face pressure to align with the group’s broader policy objectives. For businesses, this can mean different requirements for hosting, procurement, security reviews or data transfer across markets. A decision that looks inexpensive at the pilot stage can create longer-term switching costs if systems become tied to one vendor’s chips, APIs, cloud services or compliance framework.
Kazakhstan shows the balancing act
The AP report does not suggest that every country will be forced to make a clean binary choice immediately. Instead, it shows the growing difficulty of maintaining flexibility as technology partnerships become linked to national security and industrial policy. Governments will need to compare the full cost of each ecosystem: hardware availability, model performance, local skills, reliability, rules for sensitive data and the ability to move later. For smaller economies, selective participation across both systems may be possible, but only with clear limits on which technologies are used for critical functions.