Tue Jan 28 09:33:15 UTC 2025: ## Chinese AI Chatbot DeepSeek Shakes Up US Tech Industry

**Beijing/San Francisco** – A new Chinese artificial intelligence (AI) chatbot, DeepSeek, is causing ripples throughout the global tech industry, challenging the dominance of US-based AI models like ChatGPT. Developed by a Hangzhou-based research lab with a reportedly minuscule budget compared to its American counterparts, DeepSeek has rapidly gained traction, becoming the top downloaded app on the US Apple App Store.

DeepSeek’s latest models, V3 (a 671-billion parameter model) and R1 (an advanced reasoning model), are key to its success. R1, in particular, has impressed experts with its performance on logical reasoning tests, outperforming ChatGPT and Claude AI by 7-14%, and achieving a 92% success rate on complex problem-solving tasks in one benchmark. This performance comes at a fraction of the cost: while OpenAI’s GPT-4 charges $15 per million input tokens, DeepSeek’s R1 charges only $0.55 – making it 20 to 50 times more affordable.

The low cost is particularly striking considering DeepSeek V3’s training reportedly cost less than $6 million, using mid-range Nvidia H800 chips. While analysts acknowledge that this figure likely underrepresents the total cost, it still pales in comparison to OpenAI’s estimated annual expenditure of $5 billion. OpenAI CEO Sam Altman himself has called DeepSeek R1 “very impressive,” a sentiment echoed by Google CEO Sundar Pichai, who urged taking the development “very seriously.” Even former US President Donald Trump has issued a “wake-up call” regarding DeepSeek’s implications.

Furthermore, DeepSeek’s open-source nature, under the permissive MIT License, allows for unrestricted commercial use, modification, and integration, giving it a significant advantage over proprietary models. Its speed and purported superior understanding of regional idioms and cultural contexts further enhance its appeal.

The emergence of DeepSeek has sparked debate about the efficiency of massive investments in AI development by US tech giants, particularly in light of recent stock market losses in the tech sector. Nvidia, for example, saw a significant drop in its market capitalization this week.

DeepSeek’s creator, Liang Wenfeng, recently attended a closed-door symposium hosted by Chinese Premier Li Qiang, highlighting the model’s potential importance to China’s strategic AI goals. The rapid rise of DeepSeek suggests China may have significantly narrowed the gap in AI technology, challenging the previously assumed US lead. However, larger geopolitical implications and ethical considerations surrounding open-source AI remain to be addressed.

Read More