DeepSeek Review

DeepSeek is a Chinese artificial intelligence startup that has rapidly gained attention for its advanced AI models and cost-effective training methods. The company’s virtual assistant, based on its V3 model, became the top-rated free app on the U.S. iOS App Store by January 27, 2025, surpassing competitors like ChatGPT. This assistant is noted for its ability to answer questions, solve logical problems, and write computer programs, performing comparably to other chatbots in the market.

Key Features and Achievements:

• Efficient Training: DeepSeek’s V3 model was trained using approximately 2,000 Nvidia H800 GPUs over 55 days, costing around $5.58 million. This is significantly less than the expenditures of some U.S. tech giants on similar AI technologies.
• Global Impact: The competitive performance of DeepSeek’s models has been recognized as a potential challenge to the dominance of U.S. AI models. Some media outlets have referred to its emergence as a “Sputnik moment” for U.S. AI, highlighting its impressive capabilities.

Critiques and Concerns:

• Cost Underestimation: The reported budget for training DeepSeek’s V3 model excludes significant expenses such as personnel salaries, data processing costs, and system architecture design. Additionally, the hardware required for training, approximately 2,048 Nvidia H800 GPUs, is estimated to cost around $70 million, suggesting that the actual expenses may be higher than reported.
• Allegations of Data Misuse: In January 2025, OpenAI and Microsoft suspected DeepSeek of using knowledge distillation processes to extract data from OpenAI’s models via its API, potentially violating terms of use. OpenAI claimed to have evidence of these practices and indicated collaboration with the U.S. government to protect its intellectual property.
• Data Storage and Privacy: According to its privacy policy, DeepSeek stores user data, including conversation histories and personal information, on servers located in China. This raises concerns due to China’s national intelligence law, which requires companies to cooperate with authorities, potentially compromising user privacy.
• Content Censorship: Operating under Chinese internet regulations, DeepSeek’s chatbot avoids discussing politically sensitive topics such as the Tiananmen Square protests, the status of Taiwan, the 2019 Hong Kong protests, and allegations of forced labor in Xinjiang. The chatbot may redirect conversations or provide responses that align with official government positions, leading to concerns about censorship.

Industry Reactions:

The AI community has had mixed reactions to DeepSeek’s advancements. While some industry leaders have praised its achievements, others have expressed skepticism regarding its capabilities and the sustainability of its success. Notably, the emergence of DeepSeek has prompted discussions about the global AI landscape and the competitive dynamics between Chinese and U.S. AI developments.

Conclusion:

DeepSeek represents a significant development in the AI industry, showcasing China’s growing capabilities in artificial intelligence. Its efficient training methods and rapid market impact highlight both opportunities and challenges in the global AI landscape. However, concerns regarding cost transparency, data practices, and content censorship suggest that users and industry stakeholders should approach DeepSeek’s offerings with careful consideration.