"Why Drive a Ferrari When a Toyota Will Do?"... Chinese AI Makes Inroads Into Africa

Up to 90% Lower Setup Costs With Chinese AI
Kenyan Entrepreneur: "Why Choose a Ferrari When a Toyota Is Enough?"

African Developers Increasingly Turn to Chinese AI
Setup Costs Reduced by As Much As 90%, Favorable Pricing Structures
Chinese

The African information technology (IT) sector is increasingly adopting Chinese open-source artificial intelligence (AI) as its main development tool, as opposed to American-made AI. The primary reason is cost-effectiveness.

The language model "Sunflower" supports 31 indigenous languages of Uganda. It uses Alibaba's "Q-1 3" model. Official website

On August 5 (local time), Yonhap News, citing The New York Times (NYT), reported that developers in African countries such as Kenya, Uganda, and Nigeria are mainly using AI models from Chinese companies like Alibaba and DeepSeek, rather than from American companies such as OpenAI and Anthropic.

Ernest Mwebaze, an AI developer in Uganda who is a former Google research scientist, chose to build the language model "Sunflower," which supports 31 indigenous languages in his country, based on Alibaba’s "Qwen 3" after testing both American and Chinese models. He selected Alibaba’s model because of its lower costs and the ease with which it could be customized using his own data. Sunflower, based on this, provides Ugandan farmers with weather information and crop management guidance in local languages.

The decision to select American or Chinese AI models is largely determined by their pricing structures. American companies do not open their model architecture and charge users based on usage, so as the service scales, costs increase accordingly. In contrast, Chinese companies release the entire model as open source. This allows developers to download and modify it without an approval process, and as long as they have their own servers, they can operate the service without incurring additional expenses.

Moses Chemibaro, who runs a digital marketing business in Kenya, told NYT, "If a Toyota hatchback is enough to take a child to school, why use an expensive Ferrari?" He added, "Chinese AI can reduce total build costs by up to 90 percent."

The Chinese government is leveraging this trend as a pathway for AI expansion. Seven African countries have signed AI cooperation agreements with China, and Chinese President Xi Jinping stated at an AI-related event last month that the benefits of technology should not be limited to specific countries.

According to a joint report by the Massachusetts Institute of Technology (MIT) and Hugging Face, Chinese-developed open-source large language models (LLMs) accounted for 17.1 percent of global downloads in the past year, surpassing the United States (15.8 percent) for the first time. DeepSeek's "V3" and Moonshot AI's "Kimi K2" reportedly shared around 30 percent of last year's global open-source LLM usage.

Last month at the World Artificial Intelligence Conference held in Shanghai, China, a crowd gathered in front of the 'Kimi K2' booth from Moonshot AI. Photo by Reuters Yonhap News

However, there is another side to this trend. When a researcher at the independent press research institute "China Media Project" asked Sunflower questions related to China, the chatbot described China as a democratic country and avoided addressing questions about China’s economic and environmental impact on Uganda. Afterward, Chinese state media praised Sunflower, and Huawei reportedly offered Mwebaze the use of its cloud service.

However, American models are still being used for work in which accuracy is crucial, and ChatGPT and Claude are said to be widely used among Kenyan consumers. Additionally, many of the Chinese open-source models still operate on cloud platforms owned by Amazon and Microsoft, meaning that part of the value generated in data centers ultimately goes to American companies.

The NYT pointed out that "in June, after a U.S. government export control action, Anthropic's latest model 'Fable' was blocked, cutting off Kenyan users' access. This has increased awareness in Africa that overreliance on American AI is a structure that could be disrupted at any time."

This content was produced with the assistance of AI translation services.