LOGIC.co.id – DeepSeek has been making headlines recently for emerging as a formidable challenger to the dominance of Artificial Intelligence (AI) technology. In terms of format, DeepSeek is not much different from other AI chatbots.
This AI model, developed by a Chinese company, is similar to OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude. DeepSeek can respond to and process various user commands.
Although its format is similar, China’s DeepSeek has been developed more efficiently compared to AI models that have long been dominated by American companies such as ChatGPT. Moreover, with its latest model, DeepSeek has the potential to be even smarter than ChatGPT, posing a serious threat to U.S.-based AI models.
Here are some key facts about DeepSeek, which has the potential to surpass ChatGPT and other AI models developed by American companies.
Key Facts About China's DeepSeek AI:
1. Caused U.S. Tech Stocks to Plummet
DeepSeek serves as a warning to the U.S. that its dominance in AI may not last long. The launch of DeepSeek’s latest AI model on January 20 significantly impacted the stock market, particularly U.S. tech stocks.
On Monday (January 27, 2025), shares of U.S. tech companies dropped sharply. Nvidia (NVDA), the leading AI chip supplier, fell nearly 17%, losing $588.8 billion in market value—the largest single-day market value loss in history.
2. 10 Times Cheaper Than ChatGPT
DeepSeek's development has been far more cost-effective than OpenAI’s GPT-4, the model used in ChatGPT. DeepSeek-R1 was trained in about two months at an estimated cost of $6 million, whereas GPT-4’s development reportedly cost around $63 million and took several months to a year.
3. Outperforms Other AI Models in Benchmark Tests
Despite its lower cost, DeepSeek’s capabilities are impressive. It is claimed to outperform AI models like Anthropic’s Claude, Meta’s Llama, and OpenAI’s GPT in several benchmark tests.
4. Two AI Models: DeepSeek V3 and DeepSeek R-1
Since its founding in 2023, DeepSeek has developed multiple AI models under the “DeepSeek” name. The latest models, DeepSeek V3 and DeepSeek R-1, have gained significant attention.
5. Open-Source Development Model
Unlike proprietary AI models like ChatGPT, DeepSeek-R1 is completely open-source. This means anyone can use and develop its code freely, making it significantly different from the closed-source approach of ChatGPT.
6. Uses Nvidia H800 Chips
DeepSeek's lower development costs are attributed to the use of the Nvidia H800 AI chip, which is more affordable than the Nvidia H100 chips used for GPT-4o.
7. Launched Amid U.S. AI Chip Export Restrictions
DeepSeek was developed at a time when the U.S. imposed restrictions on exporting AI chips to China. The AI industry remains dominated by U.S. companies, especially chip manufacturers like Nvidia.
8. Sparked Doubts Over U.S. Tech Supremacy
According to Keith Lerner, an investment analyst at Truist Financial, DeepSeek’s emergence has led the market to question the long-standing dominance of American tech firms in the AI industry.
9. Developed Using Domestic Talent
DeepSeek was fully developed using young talent from within China. Its founder, Wenfeng, emphasized that AI innovation does not necessarily require hiring top AI experts from outside China.
10. Surpassed ChatGPT on the App Store
DeepSeek is available in multiple formats, including a mobile app. On Monday (January 26, 2025), the DeepSeek mobile app became the No.1 free app on Apple’s App Store in 111 countries.
11. Triggered a Crypto Sell-Off
DeepSeek’s emergence also impacted the cryptocurrency market. On Monday (January 27, 2025), Bitcoin’s price briefly dropped 7%, falling below $98,000 per coin.
12. Founded by Liang Wenfeng
DeepSeek was developed by High Flyer, a Chinese hedge fund, and is led by Liang Wenfeng. Born in 1985 in Zhanjiang, Guangdong Province, China, Wenfeng has played a key role in DeepSeek’s rapid growth.
13. More Resource-Efficient and Effective
DeepSeek adopts innovative AI techniques to ensure efficiency and high performance. It utilizes Mixture-of-Experts (MoE) and Chain of Thought (CoT) methods for optimization.
14. A Young AI Startup
Despite its growing reputation, DeepSeek is still a relatively young company. Founded in 2023, it is based in Hangzhou, Zhejiang, China.
15. Raised Concerns in the U.S. Government
The rise of DeepSeek has caught the attention of the U.S. government. According to Gizmochina, American authorities are closely monitoring its disruptive impact.
DeepSeek is proving to be a serious contender in the global AI race, challenging the dominance of U.S. companies and reshaping the future of artificial intelligence.