Can China Win the World Over With Open AI?

China and the United States are locked in a fierce race to lead the world in artificial intelligence. At a major AI conference held in Shanghai from July 17 to 20, China made a bold move by launching a new global AI organization and showing off powerful AI models it is sharing for free. One of the biggest draws was a new AI model called Kimi K3, made by a Chinese startup called Moonshot AI. Reporters and experts crowded around the company's booth, eager to learn how it worked. Kimi K3 can perform as well as top AI systems from American companies like Anthropic and OpenAI — and costs far less to use.

Kimi K3 is what experts call an 'open-weight' model, meaning the company shares its inner workings so anyone can download, change, and run it on their own computers. American AI companies like OpenAI and Anthropic mostly keep their models closed, so users must pay to access them through the company's servers. This difference has become a major flashpoint in the rivalry between the two countries. When Kimi K3 launched, investors got nervous and sold off shares in US and South Korean chip companies. They worried a cheaper Chinese AI could shake up the industry the way another Chinese model, DeepSeek, had done earlier in 2025.

Chinese President Xi Jinping gave a speech at the Shanghai conference — the first time he had attended since it started in 2018. He promised to work with all countries to develop AI together and share its benefits fairly. He also pledged to help developing nations with thousands of AI training opportunities and access to a Chinese AI-powered weather warning system. China also launched a new group called the World Artificial Intelligence Cooperation Organisation, or WAICO, designed to help poorer countries in Africa, Asia, and Latin America use AI safely. The United Nations secretary-general attended the founding ceremony, giving the group a high-profile stamp of approval.

The United States has its own competing plan, called Pax Silica, launched in December 2025 with 23 partner countries and the European Union. Its goal is to make sure member nations do not become too dependent on China for AI, computer chips, and other key materials. Reports say the US warned countries they could be shut out of its AI coalition if they also join China's WAICO. Only Kazakhstan is known to belong to both groups, seeing membership in each as a smart way to benefit from both sides. Many other countries, like Malaysia, are also taking a mixed approach — working with both Chinese and American tech companies at the same time.

Why would China give away powerful AI for free? One big reason is cost: Chinese open-weight models are 60 to 90 percent cheaper than closed American models, making them very attractive to budget-conscious governments and companies. Another reason is strategy — sharing AI widely puts pressure on US companies that charge for access and helps China close the technology gap faster. Technology analyst Ben Thompson has noted that China's strength in manufacturing robots and physical products means it could benefit enormously from widely available AI. By spreading its AI around the world, China also hopes to build friendships and influence with many countries.

There have been serious accusations between the two countries. The US government accused six Chinese AI companies — including Moonshot AI and DeepSeek — of secretly copying reasoning from American AI systems, a technique called 'distillation.' China called the accusations 'groundless.' Anthropic said it found evidence that Moonshot used hundreds of fake accounts to access its Claude AI and steal data from it. These tensions show how the AI race is not just about who builds the best technology — it is also about trust, rules, and who gets to decide them. Both sides are using strong language, with Chinese state media calling US tech controls an 'AI Iron Curtain.'

Security worries are adding fuel to the fire. In June, the US government ordered Anthropic to cut off foreign access to two of its AI models because of national security concerns. Countries that had been relying on those tools suddenly lost access — a wake-up call for many governments. Open-weight models, by contrast, can be run on a country's own servers, so no outside government or company can shut off access. This makes open-weight AI very appealing to countries that want to stay in control of their own technology. Experts call this goal 'AI sovereignty.'

The race is not just about technology — it is about who sets the rules for AI around the world. American AI leader Dario Amodei has called for slowing AI development and keeping democracies ahead of countries like China. President Trump disagrees with slowing down but still wants the US to win, saying: 'Whoever wins AI, wins.' Meanwhile, African nations are watching carefully, since the continent receives only about 0.3 percent of global AI investment. African researchers say they want tools that truly help their countries grow, not just serve the interests of bigger powers.

The real picture emerging is messier than a simple US-versus-China split. Big American tech companies like Microsoft, Amazon, and Google are reportedly in talks to offer China's Kimi K3 on their own cloud platforms. Countries are choosing AI tools based on cost, need, and relationships rather than loyalty to one side. Kazakhstan belongs to both WAICO and Pax Silica, and Malaysia works with both Chinese and American companies. The AI race is intense, but the world is not simply choosing sides — it is navigating a complicated mix of politics, money, and technology.