Kazakhstan broke the unwritten rules of this new technological cold war. It signed the "AI Opportunity Statement," joined Pax Silica, and also entered the World Artificial Intelligence Cooperation Organization that Xi Jinping launched in July. Washington responded with an ultimatum that, in diplomatic terms, reads as elegant. You cannot play at two tables. Thirty-five countries are about to receive the same message.
Pax Silica is a US-led alliance that seeks to control access to critical minerals, semiconductors, and frontier AI models by making economic cooperation conditional on geopolitical exclusivity. The name evokes a peace treaty. The reality looks more like a siege. Peace here doesn't mean the absence of conflict, but centralized dominance disguised as stability — the same rhetorical device empires used when naming the Pax Romana or the Pax Britannica.
The prevailing thesis has internal logic. The United States needs to secure lithium, cobalt, rare earths, and chip manufacturing channels before China consolidates its own alternatives. If dozens of countries can belong to both blocs simultaneously, Pax Silica loses its leverage. Any ally could exploit both sides at once. Forcing Kazakhstan to choose thus responds to a strategic survival need for the alliance.
The argument rests on a real asymmetry. Dependence on critical minerals is not balanced. Kazakhstan produces uranium, chromium, and other inputs that neither the United States nor China can easily replace. When a supplier splits loyalties between blocs, no alliance reaches the monopoly it's pursuing. Strategists in Washington didn't invent this problem. They already lived through what dependence on a vulnerable source means with Taiwanese semiconductors. They want to avoid repeating it, now from an offensive position.
Open Chinese models add further weight. Alibaba's Qwen family and DeepSeek's systems are being distributed at a pace that unnerves OpenAI and Anthropic. A model that can be downloaded for free, adapted locally, and deployed without US infrastructure erodes Silicon Valley's advantages — not because of inferior technology, but through sheer accessibility. This is pure, hard-nosed market calculation.
Why should Kazakhstan accept picking just one side? The official answer talks about security and chains of trust. The real answer is cruder. Both the United States and China benefit from fragmenting the map into closed blocs. That fragmentation maximizes their influence over economies negotiating from a structurally disadvantaged position. Those who hold the resources end up turned into bargaining chips.
An earlier article on the state-level convergence between Washington and Beijing in dominating frontier AI already pointed this out. Both governments agree that advanced AI must remain under centralized state oversight, whether through sanctions and exclusive alliances or through licensing and direct regulation. The competition between Pax Silica and the Chinese organization doesn't pit an open model against a closed one. These are two versions of the same closure, each with a different owner.
This places countries like Kazakhstan in the historic position of the voiceless supplier. Their minerals become geopolitical currency while real power stays with whoever processes, manufactures, and distributes. Resource dependency analyses have shown this constant for decades. Whoever owns the raw material rarely dictates the final terms.
"Us versus them" coalitions grow stronger when a shared external threat appears. Jane Goodall documented this dynamic among chimpanzees at Gombe. Pax Silica isn't reacting to China alone. It uses China as a necessary enemy to justify exclusive loyalty. Countries that would prefer diversified relationships watch that option close off. The message to Kazakhstan normalizes the idea that neutrality is no longer acceptable.
A third way exists, though both blocs would rather it went unmentioned. Lightweight AI models trained locally, with distributed governance and no exclusive dependence on any superpower, are already technically viable. They don't require OpenAI's scale or Chinese state backing. The obstacle is political. Neither Washington nor Beijing has any interest in funding what escapes their centralized control.
Stones Don't Lie explores how Greek cultural hegemony operated without direct conquest. It was enough to make alternatives seem uncivilized. Pax Silica applies a similar logic, though with less subtlety. It doesn't invade. It simply establishes that there are only two legitimate paths toward the future of AI, and both demand exclusivity.
Kazakhstan didn't sign both agreements out of confusion. Dual membership was perfectly rational under the rules in force at the time. The international system allowed it for months. The rule it supposedly broke was invented afterward, to correct what had previously been acceptable. Dual membership. Rational then.
I still don't know how this tension will be resolved without suppliers ultimately losing their autonomy. I'm still working through this. What if the real choice weren't between Washington and Beijing, but rejecting outright the need to pick a side in this silicon war?
