The rise of China’s AI! U.S. stock futures and Bitcoin all fell: What is the difference between DeepSeek and ChatGPT?

Here is the English translation of the text, with the specified terms preserved: The S&P 500 futures in the US fell 1% to 2% today, raising investors' concerns about the high valuation of tech stocks. At the same time, some China-related stocks bucked the trend and rose, further deepening the market's thinking about the technology competition between China and the US. The impact of China's low-cost AI industry has led to a revaluation of the overall market, and this market volatility has also made a new Chinese AI company, DeepSeek (Hangzhou Shendu Qiusuo), a global focus of attention.

China's AI Rising Star: DeepSeek's Low-Cost Innovation Shocks the US

DeepSeek is an artificial intelligence system developed by a Chinese company, which has emerged globally with its open-source large-language model and extremely high cost-effectiveness. DeepSeek's latest model, DeepSeek-R1, was launched on January 20, 2025, with a total cost of only $6 million. In contrast, the development cost of mainstream US AI models like OpenAI's CHATGPT reaches tens of billions of dollars. This low-cost but high-performance model is not only highly acclaimed in the Chinese domestic market, but has also attracted great attention in the international market. The founder of DeepSeek was recently invited to meet with Chinese Premier Li Qiang, and Chinese domestic media such as Xinhua News Agency have also given positive coverage of its achievements. This company is gradually becoming a focus of heated discussion at home and abroad.

DeepSeek vs. CHATGPT: Comparison of the Two AI Giants

Although DeepSeek and OpenAI's CHATGPT have obvious differences in design concept, technical path and market positioning.

DeepSeek's Expert Mixture Model Challenges the US Hardware Advantage

DeepSeek's development is oriented towards the needs of the Chinese market and national policies, focusing on the combination of high performance and low cost. It utilizes limited hardware resources to achieve performance close to global leading technologies. CHATGPT, developed by the US OpenAI, aims to create a multi-functional general-purpose conversational AI. CHATGPT is based on massive datasets and uses cutting-edge hardware for training, with the goal of providing a wide range of applications and precise user experience. In terms of hardware and cost-effectiveness, there are also differences. DeepSeek relies on the innovative technical architecture of "Mixture of Experts" to significantly improve computing efficiency under limited hardware resources. Its development cost is only $6 million, which can be considered a model of high cost performance. OpenAI, on the other hand, has invested billions of dollars in model development and training, relying on cutting-edge GPU and TPU hardware to ensure the high performance and multi-functionality of the model.

Different types of content restrictions for DeepSeek

• DeepSeek: Restricted by local Chinese regulations, its model avoids discussing sensitive topics such as the Tiananmen Square incident or Chinese government policies, but may be more critical of other countries.
• ChatGPT: Follows OpenAI's ethical principles, striving for neutrality and avoiding content that could be harmful or politically biased, but enjoys a wider range of discussion globally.

DeepSeek's strengths differ from GPT

• DeepSeek: Excels at reasoning-related tasks and has even surpassed or approached ChatGPT in some benchmark tests. Its model focuses on resource optimization and high-performance.
• ChatGPT: Emphasizes conversational ability, multilingual support, and broad adaptability, suitable for education, entertainment, and enterprise solutions.

DeepSeek can be positioned for the mass market

• DeepSeek: Primarily targeting the Chinese market, but its low-cost, high-performance strategy is rapidly gaining global attention, becoming a formidable challenger to Western AI companies.
• ChatGPT: Oriented towards the global market, emphasizing diverse application scenarios, covering education, enterprise, and individual users.

Innovation behind the low cost: DeepSeek's technical advantages

The key to DeepSeek's ability to develop high-performance models at low cost lies in its technological innovation. Its "expert ensemble model" is a revolutionary AI architecture that can significantly reduce computational resource requirements. This not only reduces hardware dependence, but also helps DeepSeek maintain competitiveness even when facing US export restrictions (such as the NVIDIA chip ban).

Furthermore, the early accumulation of high-end chips by Chinese tech giants like Alibaba and Tencent has also indirectly supported small innovative companies like DeepSeek. In the global ranking of large language models, DeepSeek has firmly established itself in the top ten, representing the progress of China's AI technology.

China-US AI competition: Proving China's strength

The rise of DeepSeek once again demonstrates China's strong capabilities in AI software innovation, especially in the face of limited resources. In contrast, US companies like Microsoft and OpenAI rely on high-cost hardware solutions. However, analysts point out that US AI companies may emulate DeepSeek's technical architecture and adopt similar "expert ensemble models" to reduce costs and narrow the gap.

It is worth noting that China currently has the largest number of software developers globally, about three times that of the US. This fact suggests that Chinese AI companies may continue to maintain a strong momentum of innovation in the future competition.

(Stargate Advances US AI Infrastructure! Microsoft CEO: Continuing to Invest in the AI Ecosystem, Collaborating with OpenAI to Stabilize)