Washington's fixation on frontier AI models is blinding it to a more immediate and consequential battlefield: the developing world. While U.S. policymakers debate the latest breakthroughs in generative AI, China is quietly winning the commercial race across Africa, Asia, and Latin America by offering affordable, adaptable technology that governments and businesses can actually use.
Most of the world doesn't need the most advanced AI system. They need something that works, is cheap, and can be integrated into local infrastructure. China has understood this. Its open-weight models, such as DeepSeek and Alibaba's Qwen, are freely downloadable and customizable, unlike many American models that are locked inside proprietary cloud services. This openness lowers the barrier to entry and encourages local developers to build on Chinese technology.
The stakes are high. The countries that choose Chinese AI today will likely build their digital infrastructure—hospitals, classrooms, airports, and government services—around it. Once that happens, switching costs become prohibitive, and Beijing gains a long-term strategic foothold. This is not just about commerce; it's about influence. A government that relies on Chinese cloud computing and technical support may find it harder to challenge Beijing on other issues.
Huawei is already expanding its footprint. In April, it launched AI services in nine new countries, including Brazil, Indonesia, and South Africa, offering access to DeepSeek and other Chinese models. In Egypt, Huawei provided eligible startups with up to $150,000 in cloud credits, training, and an Arabic-language model. These incentives are designed to seed Chinese technology into local ecosystems.
Data shows the strategy is working. Microsoft estimated that DeepSeek usage in Africa was two to four times higher than in other regions in the second half of 2025. Chinese cloud providers now control significant market shares in Latin America: 25% in Brazil, 40% in Chile, and 100% in Argentina, Mexico, and Peru. In Southeast Asia, Chinese firms own 57% of Indonesia's data center market and 100% in Thailand, the Philippines, and Malaysia.
There are also security risks. U.S. government testing found that DeepSeek models repeated Chinese Communist Party narratives more often than American models. As people increasingly turn to AI for information, Beijing's preferred version of history and current events could gain wider acceptance. Purchasing Chinese technology also raises concerns about surveillance and data security, which could undermine democratic values and national security in adopting countries.
The U.S. response has been piecemeal. The Trump administration has promoted American AI packages and expanded export financing, but these efforts are insufficient. The dismantling of USAID and the lack of a coordinated strategy for the Global South have left a vacuum. As Pope Leo's AI warning highlights, technology should serve humanity, not become a tool of geopolitical coercion.
To compete, the U.S. must significantly increase funding for countries interested in purchasing American AI, invest in training local users and developers, support tools in local languages, and work with partner governments to solve their specific problems. The U.S. also needs to counter Chinese influence operations, as OpenAI's recent bans on foreign influence networks show the importance of protecting the information space.
China is ahead in the Global South, and the U.S. is losing the broader AI race. Winning back these markets is not just an economic imperative but a strategic one. If Washington fails to act, it will not only lose sales but also its ability to shape how AI is developed and used worldwide. The consequences for U.S. influence and national security would be severe.
