Deepseek vs ChatGPT: Can Africa follow suit in the next AI revolution?

The launch of DeepSeek, a Chinese-developed artificial intelligence (AI) model, has sent shockwaves through the generative AI landscape and the broader tech ecosystem. Its latest iteration, DeepSeek-R1, quickly soared to the top of Apple’s App Store downloads, fuelling speculation about its disruptive potential. Its emergence rattled US markets, triggering a decline in major tech stocks, including NVIDIA and Nasdaq-listed firms.
Reportedly, DeepSeek-R1, which is as efficient as OpenAI’s ChatGPT, was developed at a fraction of the cost. While OpenAI CEO Sam Altman hinted at spending $100 million to build ChatGPT-4, DeepSeek-R1 was created for just $6 million.
Democratisation of AI
While speculation remains on the accuracy of this figure, DeepSeek-R1’s emergence challenges long-held assumptions about AI development costs, especially in the face of US regulatory and technological barriers. United States tech giants have long been assumed to hold an AI edge due to their scale, talent acquisition and vast financial resources. However, DeepSeek challenges this notion. OpenAI, founded a decade ago, has 4,500 employees and has raised $6.6 billion in funding. DeepSeek, which is less than two years old, operates with only 200 employees and was developed for under $10 million.
Could its success signal the democratisation of artificial intelligence? And does this mean countries in the Global South now have a real shot at securing their place in the AI revolution?
Lessons from DeepSeek’s AI disruption
DeepSeek’s rapid emergence as a ChatGPT competitor proves that AI innovation is not exclusive to Silicon Valley. Its success offers a playbook for Africa’s AI ambitions: building high-performing models with limited resources, overcoming hardware restrictions and fostering homegrown AI solutions.
DeepSeek demonstrated that cutting-edge AI does not require the latest GPUs. It trained its model using 2,000 NVIDIA H800 chips, which are less advanced but still effective. African AI startups can adopt similar efficiency strategies by optimising AI models for lower-end hardware. Governments and businesses should invest in AI research that prioritises computational efficiency rather than dependence on high-end imports.
DeepSeek’s success also stems from operating with just 200 employees and a development cost under $10 million, compared to OpenAI’s massive workforce and $6.6 billion in funding. Africa can create low-cost, open-source AI models tailored to local needs, avoiding the high costs associated with Western AI labs.
China has built a strong AI ecosystem by investing in local talent and research institutions. DeepSeek benefits from this approach, proving that a well-trained workforce can build world-class AI models. Similarly, organisations like Samasource, a business process outsourcing company that trains young Kenyans in data labelling and annotation for AI, demonstrate the continent’s potential in the global AI value chain. Universities and governments must fund AI research programmes to train African data scientists and engineers. AI training should focus on localised solutions, including natural language processing (NLP) for African languages to ensure digital inclusion.
DeepSeek represents China’s effort to reduce reliance on Western technologies, ensuring AI and, ultimately, data sovereignty. With the recently approved Kenya Cloud Policy emphasising data sovereignty, DeepSeek proves that emerging AI ecosystems can challenge dominant players. With the right investments, policies, and focus on AI efficiency, Kenya, too, can develop its own formidable ‘DeepSeek.’