Kazakhstan has recently signed up for two rival AI coalitions, one from the US and the other from China. It joined the American Pax Silica initiative to secure supply chains in critical minerals, semiconductors and artificial intelligence on June 25, 2026.
It was also one of the 29 founding members of the World Artificial Intelligence Cooperation Organisation, which was set up in Shanghai on July 16 as an intergovernmental organisation. Kazakhstan’s decision may offer an early indication of a challenge Washington will increasingly face in the global AI race: countries may be willing to align politically with the US while continuing to engage technologically with China.
On August 14, Reuters published a letter that was reportedly drafted by the US State Department and addressed to the 35 signatories of Washington’s AI Opportunity Statement. The draft said Pax Silica membership is incompatible with membership in duplicative initiatives with expectations that conflict with those of the US, and it encouraged countries to make a conscious choice. The State Department declined to comment on the leak, but an anonymous US official told Reuters that a country “can’t have it both ways”.
The US has a precedent for pressuring partners to exclude Chinese technology as it did in its campaign against Chinese telecommunications companies. However, AI poses a different enforcement challenge. A Chinese AI model doesn’t need to be physically transported across a border. It can be downloaded, copied, modified or deployed locally.
5G hardware was physical, concentrated in identifiable networks and thus relatively auditable. AI is much more pervasive. A government can sign Washington’s declaration and pledge to phase out Chinese AI, while its institutions, companies or researchers can continue to use Chinese models. The problem is not that Chinese AI leaves no footprints. The issue is that those traces are disjointed and frequently beyond Washington’s control.
Washington can address this issue by regulating the technologies that enable AI, not just the models themselves. On August 4, Reuters reported that the Trump administration was drafting a ban on new models of Chinese optical transceivers used in data centres. At the other end of the technology spectrum, Washington has been pressuring the Netherlands to impose stricter limits on the sale and servicing of chip-making equipment, which is critical to producing advanced semiconductors used in AI systems. The Dutch government has raised objections to proposed US legislation that would further restrict exports of chip-making equipment to China.
This exposes another issue. The wider the scope of the technologies that countries must avoid, the more expensive it is for Washington’s partners to align. While optical transceivers are less technologically advanced than cutting-edge chip-making machines, both illustrate the same US approach centred on limiting Chinese access to AI and the broader technological ecosystem that enables it.
The National Security Science and Technology Strategy released by the White House on August 18 points to tools including investment restrictions and export controls that Washington can use to shape the global technology landscape. These measures could enhance US leverage, but they also impose more requirements on partners whose economic interests are not necessarily aligned with Washington’s.
Traditional US allies like Japan, Australia and South Korea may be better equipped to absorb the costs of alignment because of their technological capabilities and access to US-led technology ecosystems.
The challenge is more difficult for many developing countries. Their technological ecosystems often rely on both Chinese infrastructure and cooperation and American semiconductors, cloud services and other technologies. They cannot easily give up Chinese technology, and they risk losing access to American technology if they get too close to Beijing. In these countries, choosing between US- and China-led AI ecosystems is not just a political decision. It can involve substantial economic and technological sacrifices.
This vulnerability is reflected in China’s response on August 19. Beijing called for respect for the “digital sovereignty” of every country and opposed the idea of opposing camps in AI, saying states should choose partners based on their national conditions and development needs. China can also offer countries incentives to expand AI cooperation. This argument might appeal to countries, including Kazakhstan, that are not prepared to accept binary technological alignment.
Interestingly, Kazakhstan is a key potential source of critical minerals, which constitute a major component of Washington’s AI coalition. Yet China also retains an advantage here. China demonstrated its leverage over rare-earth supplies in 2025 during its trade conflict with the US. So, until there is an alternative, threatening to cut off mineral-rich partners such as Kazakhstan also threatens to cut off partners Washington may least want to lose.
Washington’s dilemma is thus not just that states may be able to participate in two competing AI alliances. The measures that proved effective against Chinese telecommunications companies may not be sufficient to secure full technological alignment in AI. Although the US can restrict access to chips, infrastructure and investment, AI models themselves are considerably harder to control across an entire national ecosystem.
Kazakhstan’s dual membership in US and Chinese-led AI blocs thus highlights how difficult it will be for Washington to force other countries to choose between American and Chinese AI. The US challenge is not merely to secure that choice, but to verify compliance.
The writer is a research assistant at the Centre for Aerospace & Security Studies (CASS), Islamabad. He can be reached at: [email protected]
Disclaimer: The viewpoints expressed in this piece are the writer’s own and don’t necessarily reflect Geo.tv’s editorial policy.
Originally published in The News
2026-09-17 09:18:00
