95점짜리 중국 AI, 비용은 6분의 1입니다 - 김덕진 IT커뮤니케이션 연구소 소장

#AI #USA #BigTech #Microsoft #China #Hegemony #Nadella #Ecosystem #Frontier #Data #DeepSeek #V4 #OpenAI #MS #Antropic #Google #Cloud #Price #Huawei #XuZhijun #Security #HongKong #ZhiFu #Claude #DokFaMo #SovereignAI #Upstage #ParkJungHo #ProfessorParkJungHo [Content summarized via AI] 📝Summary: AI Hegemony, Now a War of 'Cost-effectiveness' Over 'Performance' It explains that the AI ​​competition between the United States and China has entered a new phase. While the key in the past was creating AI models with the highest performance, the ability to use a large number of AI agents cheaply is now emerging as a competitive edge. Amidst this shift, Microsoft is considering adopting China's DeepSeek models into its AI services, and even companies that have made large-scale investments in OpenAI are showing signs of utilizing Chinese models to reduce costs. This demonstrates that the AI ​​market is being reorganized around 'efficiency and cost' rather than 'highest performance.' DeepSeak is able to achieve low prices thanks to its efficient Architecture of Excellence (MoE) and a design optimized for Huawei's Ascend chip. U.S. semiconductor sanctions have actually spurred China's independent technology development, and coupled with active support from the Chinese government, its price competitiveness has significantly increased. While its performance falls somewhat short of the top U.S. standards, it is considered sufficient for repetitive AI tasks, attracting the attention of global companies. However, security remains the biggest variable when using Chinese AI. Due to China's National Information Law, there are concerns that data could be transferred to China if Chinese servers are used directly. In response, Microsoft is attempting to resolve security issues by downloading only the models and running them on its own cloud. Nevertheless, dependence on the Chinese ecosystem and model bias remain challenges. Meanwhile, the U.S. is strengthening a closed strategy centered on government approval by restricting the disclosure of the latest AI models. Conversely, China is expanding its ecosystem by actively releasing Open Weight models that anyone can use. Furthermore, despite massive revenue growth, OpenAI faces growing concerns regarding profitability due to high operating losses and the burden of future large-scale computing investments. It is emphasized that while Korea is also pursuing the development of its own AI models, securing global competitiveness requires more than just creating models; it necessitates the combination of diverse practical applications that companies can utilize and strong government support. ⏰ Timecode 00:00 AI Market Shifts from Performance to Cost Competition 03:47 DeepSeak's Low-Cost Strategy and the Role of Huawei Chips 07:49 Security Controversies and Application Methods of Chinese AI 11:16 Diverging Directions of US and Chinese AI Ecosystem Strategies 13:21 Round 2 of AI Hegemony and China's Offensive 15:30 OpenAI's Financial Burden and Challenges for Korean AI 🎯 Key Summary ⚡ The outcome of the AI ​​competition is now shifting from 'highest performance' to 'how much can be utilized at a low cost.' ※ Host: Park Jung-ho, Professor at Myongji University ◈ Broadcast Schedule YouTube: Daily 10:30 AM – 11:20 AM Radio: MBC Standard FM Daily 10:05 PM – 11:00 PM