AI Roundup: DeepSeek- Discover the Chinese AI That's Disrupting the Market

DeepSeek, a project by a Chinese hedge fund, High-Flyer, is currently dominating conversations in AI and tech circles. Designed like OpenAI's ChatGPT, DeepSeek was reportedly trained for "just US$6 million," compared to the $100 million funding for OpenAI's GPT-4 in 2023 and approximately one-tenth of Meta's budget for LLaMA 3.1.
On January 10th, DeepSeek was introduced to the general public without match fun fare. It's a free AI assistant, much like the hugely successful OpenAI's ChatGPT. It was advertised to use less data and was developed at a fraction of the cost of current AI models. After a few days in the market, DeekSeek surged in popularity among US and global users and even became the top-downloaded AI chat assistant app online. Then, something happened to tech stocks and the broader market.
On January 27th, global investors dumped tech stocks over worries that DeepSeek would threaten the industry leadership of traditional leaders like NVIDIA. Investors' fear and its impact on the market were swift and astounding: NVIDIA shed $593 billion from its market cap, and other tech companies suffered the same fate. So, what is DeepSeek, and why is it causing buzz in the AI industry?
What is DeepSeek?
DeepSeek R1 is a new AI model that's currently disrupting the AI industry and for the right reasons. It's an AI chat model that works like ChatGPT since the users can input prompts and questions, which it can process and answer within seconds. Early reviews say that DeepSeek can match the capabilities of the more popular AI models like Google's Gemini and OpenAI's GPT-4.
More importantly, DeepSeek is "way cheaper", with its development team claiming that the project only costs $6m (£4.8m) to train, which is just a fraction of the reported $100 million budget spent on GPT-4.
Reports also indicate that the DeepSeek project minimized the US trade restrictions on AI chips. DeepSeek's founder secured "just enough" Nvidia A100 chips, even though these high-value products were banned from export to China since September 2022. Some experts say that the development means managed to combine these NVIDIA chips with cheaper and less sophisticated ones, which surprisingly produced more efficient processing.
This development "is shocking" for other US AI and tech companies, considering that the goverment has been trying to "restrict the supply of AI chips to China" for security concerns in the last few months. Interestingly, High-Flyer's AI unit announced through its official WeChat account that as of July 2022, the company "owns and operates a cluster of 10,000 A100 chips".
This Chinese AI chat model also reportedly relies on less memory, ultimately making it a more cost-efficient AI chat model. DeepSeek's reported fast performance and cost-efficiency pushed it to become the most downloaded free app on Apple's App Store during its first week of release in the United States.
Know the company behind DeepSeek
DeepSeek is based in Hangzhou, China, with Liang Wenfeng as its controlling shareholder and co-founder of High-Flyer, a Chinese hedge fund. Although High-Flyer was established in February 2016, it was only in March 2023 that Wenfeng announced an AI project outside its traditional interest in trading.
On its official WeChat account, Liang's Hedge fund announced that it's "starting again" to concentrate its resources on creating a new and independent research group, specifically focusing on Artificial General Intelligence (AGI). OpenAI, ChatGPT's developer, defines the AGI as an autonomous system that can exceed human's capabilities
DeepSeek was launched a year later and released its AI models in November 2023. In November 2023, the company released its first model, the DeepSeek Code, and on November 29th, it added the DeepSeek LLM series of models.
In January 2025, the company released its first free DeepSeek chatbot, based on the DeepSeek R1 model for Android and iOS devices. By January 27th, 2024, the app surpassed ChatGPT as the most downloaded free app on the Apple Store in the US, and this surprising result caused NVIDIA's share to sink by 18%, dragging the other US tech stocks along the way.
Understanding DeepSeek's impact on the growing generative AI industry
Although DeepSeek's launch has caused major tech companies to shed some value off their market cap, many say its arrival is good for the industry. China's remarkable progress in the industry has helped popularize the technology and push for its wider maker adoption. If OpenAI's ChatGPT showed the capabilities of generative AI, which can change how we work, then China's DeepSeek successfully demonstrated that it's possible to deliver the same results at a fraction of the cost.
According to Andrew Yao Chi-chih, a Turing Award winner from Tsinghua University, DeepSeek can become more beneficial for the community in the long term since it's an open-source project.
Also, DeepSeek's breakthrough has created excitement among local Chinese companies. Many companies have announced that they're now using the technology to streamline their operations. According to SCMP, Lenovo, Shenzhen-based firm UBTech, and Geely were among the local companies that used and integrated DeepSeek's AI model into their products.