At the start of this year, AI experts and the tech world were snug in the knowledge that AI and technology will drive development into the next century and possibly beyond – and they are correct. Leading technology companies are betting big on AI, underscoring its huge potential to transform the world. For instance, Meta announced it was investing about $65 billion in AI development in 2025 while Microsoft promised to spend $80 billion on AI in its current financial year.
Enter Chinese DeepSeek
However, AI development and growth, many believed, would be driven by the Global West, and until this month that had been the case, especially as American export restrictions forbid the sale of advanced AI chips or semiconductors to China or Chinese interests, North Korea, Russia, Iran, and other countries considered a threat to the free world to hinder their AI development. On January 20, a little-known Chinese firm Deep Seek threatened to upset the balance of power in the AI world by releasing China’s version of artificial intelligence large language models (LLMs), the DeepSeek R1.
How DeepSeek is redefining AI norms
The release of the Chinese version of OpenAI’s ChatGPT is causing disquiet in the AI and tech world. To use a popular slang, DeepSeek has been shaking industry tables since its release. For starters, unlike the poor attempt by Chinese search engine giant Baidu to introduce the first Chinese version of ChatGPT, DeepSeek is as good, if not better, as ChatGPT and other AI chatbots, despite restricted access to cutting-edge semiconductors due to the export restrictions. OpenAI’s Sam Altman and other AI experts have praised DeepSeek’s quality. “DeepSeek’s R1 is an impressive model, particularly around what they’re able to deliver for the price,” Altman said. Talking about price, the company said it spent about $6 million on its AI model, making it far cheaper than the spend on ChatGPT -4 and other Western versions. It also requires lower computing power. This has called into question the humongous AI spend by Microsoft, Meta, Nvidia, OpenAI, and others. Most interesting is the fact that DeepSeek is open source, making its source code freely accessible to anyone to view or modify to build their own generative AI chatbots; it’s like building a complex structure and then granting free access to the blueprint to whomever is interested in building similar structure or wishing to improve on your structure. Put plainly, DeepSeek is undercutting the West’s monopoly in AI.
DeepSeek is muddling the AI investment space too
The quality, functionality, low cost, and other benefits of DeepSeek are too apparent to ignore. It was no surprise therefore that the app became the top-rated free application on Apple’s App Store in the US. A bigger industry disruption by DeepSeek, however, is the doubt it sowed in investors’ minds. Investors are worried about their huge investments in AI stocks and expected ROI. They made such investments based on market exclusivity, which is reflected in AI stocks’ high valuation. DeepSeek has however challenged the West’s AI exceptionality and the huge capital expenditure on generative artificial intelligence chatbots and investors are nervous and are asking questions or cutting back on AI and tech investments.
Carmignac Kevin Thozet, for instance, revealed that post-DeepSeek, he had cut his exposure to US tech stocks. Last Monday, retail investors and traditional asset managers dumped circa $600 billion worth of Nvidia shares as DeepSeek gained popularity. The reasoning was simple, if DeepSeek can develop a cost-effective, high-quality AI model with limited access to Nvidia’s high-end AI chips, then the AI chipmaker may not be as dominant as believed and may in fact struggle, in the face of such cheap competition, to generate high ROI.
Key lessons for Africa and other regions
Africa has a lot to learn from the unfolding DeepSeek drama. Africa has always taken the back seat in innovation and technological development. In AI development, for instance, Africa has remained largely anonymous, despite its recent gains in the tech industry through a vibrant start-up ecosystem, strong advocacy for ethical AI adoption, and growing home-grown tech talents. Infrastructure remains a huge challenge so are mass migration and a poor STEM ecosystem that is birthing a dearth of AI professionals, and limited access to AI and other tech tools.
DeepSeek fought against some of these shortcomings. It is clear that African countries, particularly Nigeria, must quickly move from developing AI strategies to implementing them. Such strategies must of necessity include the identification, support for, and development of African talents. The continent is already behind in the AI space and must act now if it hopes to benefit from the opportunities in AI. DeepSeek has shown it can be done with limited resources and without the support of the traditional tech giants.
DeepSeek versus ChatGPT
As a tech player and enthusiast, my curiosity was piqued as AI and other tech experts swooned about DeepSeek. So, I signed up for ChatDeepSeek and pit its functionality against ChatGPT and what I found was quite interesting. I posed the question “IS JAPA GOOD OR BAD FOR NIGERIA’S ECONOMY?” to the two chatbots. Both were fast with their responses. Both segmented their answers, listing the benefits and the drawbacks of japa, and both had opening and closing paragraphs. And that’s where the similarities ended. DeepSeek was more detailed with a 461-word response compared to ChatGPT’s 236-word response. DeepSeek demonstrated a better understanding of the term japa and the trend as it first explained the origin of japa. ChatGPT uses emojis to highlight its points and in its conclusion throws back my question at me.
DeepSeek has laid a marker and has shown what is possible even with limited resources. It would be interesting to see how the AI and tech industry in the West respond to this to defend its market and ensure it maintains its dominance in the industry.
Join BusinessDay whatsapp Channel, to stay up to date
Open In Whatsapp
