The artificial intelligence (AI) sector is one of the most competitive areas of contemporary technology. Developing new models requires huge investment, computing power and top experts, and any innovation can upset the balance of the market. For many years, the US giants dominated this space. Chinese AI companies, such as DeepSeek, are coming up with models that outperform the competition and offer solutions at a fraction of the price. This development challenges the American dominance in and also the huge investments that big tech companies are putting into its development.
DeepSeek has caused a stir
Most recently, a new AI R1 model from Chinese startup DeepSeek has stirred the waters of the technology sector. This open-source model has caused a revolution by relying on a lower-performance line of chips from Nvidia. The latest model, along with the one introduced just a short time ago, managed to outperform the large language models of the biggest US players. Independent third-party tests, according to CNBC, showed that in a wide range of tasks they were better than Meta's Llama model, the Claude Sonnet developed by Anthropic, or than the best-known ChatGPT from the OpenAI. That one, along with Google's Gemini model, was also surpassed by DeepSeek R1 in the ranking of the most downloaded free apps in the U.S. on the Apple store as of January 27, 2025. According to the Chinese company, they managed to make such a development progress in about 2 months, with an investment not exceeding USD 6 million. For comparison, for using one million DeepSeek R1 tokens, a user pays CNY 16 (USD 2.2), in the case of ChatGPT it is CNY 1,438 (USD 198).[1] Meanwhile, this amount has caused a wave of uncertainty in the market, as its competitors are spending several times more on AI infrastructure. Crucially, DeepSeek has also managed to develop the model despite export restrictions on advanced semiconductors by the US.[2] This is made possible largely due to the close collaboration of Chinese experts, which, according to the BBC, also lowers the cost of the products.[3]
Recognition received from leaders
US President Donald Trump has commented on the new product on the market, calling it a "wake-up call" for US companies to be inspired and deliver the same performance for less money.[4] Positive reactions were also offered by competitors. For example, the CEO of industry leader Nvidia, Jensen Huang, praised the company for its great progress[5], while the founder of OpenAI, Sam Altman, called it impressive and is happy for the new competition.[6] Microsoft CEO, Satya Nadella, referred to it as computationally efficient and warned that Chinese innovation should be taken seriously. Similar rhetoric has been adopted by Japanese minister Masaaki Taira, who, according to France24.com, said that China is not as far behind in AI development as had been thought.[7]
There are more such models on the Chinese market
Another company making rapid progress in AI is ByteDance, which is positioning itself as a strong competitor with its own high-end models. The company, which is widely known thanks to the TikTok social network, has unveiled the Doubao-1.5 Pro language model. Like DeepSeek, it outperformed its competitors ChatGPT and Claude Sonnet in certain areas, according to Medium, while also being affordable.[8] In addition to ByteDance, Baidu and Tencent are both making significant strides in this area. Baidu, famously has improved the generative AI in its Wenku platform. Thanks to it, the Chinese company has managed to significantly increase its user base as well as revenue. Tencent, in turn, is focusing on incorporating AI agents into its WeChat system.[9]
Global stock market falls
The market reaction to the release of DeepSeek R1 has been tremendous and the technology sector across the globe saw a sell-off on January 27, 2025. Among the biggest losers were giants Nvidia and Broadcom, which plunged 17% that day. Nvidia closed at USD 118.42, wiping about USD 600 billion off its market capitalization, and Broadcom fell to USD 202.13.* The NASDAQ Composite Index also dropped 3%, to 19,341 points.* By Tuesday, January 28, 2025, after the market opened, both companies and the index were already in the green. Nvidia rose to USD 121.34, Broadcom's shares were priced at USD 204.31 and the NASDAQ exceeded 19 570.* The shares of European companies in this sector also saw corrections at the beginning of the week, as did companies listed on the Tokyo Stock Exchange. Stock exchanges in the remaining Asian countries were closed due to holidays.[10]
Nvidia's share price development over the last 5 years. (Source: Investing.com)*
Broadcom's share price development over the last 5 years. (Source: Investing.com)*
The price performance of the NASDAQ Composite Index over the last 5 years. (Source: Investing.com)*
Wider adoption of AI
China's AI sector is developing at a pace comparable to that of the U.S., as evidenced by the aforementioned news. This is due not only to massive investment, but also to consumer demand. They are increasingly using products from Chinese brands that already have AI features integrated into them and are better suited to the local market. Regulations remain an obstacle, but this is not so difficult when used in applications. The challenge is the adoption of AI in the corporate world, which is still rather reticent for the most part, mainly due to legal and copyright regulations. But even here the trend is slowly starting to change. Data from analyst firm Gartner, says that in about six months, the number of companies using AI in China has increased to 10%, up from 8% previously. AI from China is expanding its reach globally, as seen in Alibaba's Accio search engine, which streamlines search for international businesses.[11]
Conclusion
The rise of Chinese AI suggests that technological dominance may no longer remain the preserve of US companies. DeepSeek proves that China can compete not only on price, but also on technological maturity. Nevertheless, it remains questionable whether these models will actually be able to outperform giants such as OpenAI, Google or Anthropic in the long run, not only in specific tasks but also in broad commercial deployment.
*Past performance is no guarantee of future results.