DeepSeek is defined as a Chinese artificial intelligence (AI) model in a simple way, but it is not so simple and has taken the tech world by surprise. It has quickly become the most downloaded app on Apple’s App Store. The AI chatbot was released on January 20 and since then has impressed experts as well as shaken up financial markets. The model has gained attention for its capabilities and for its low development cost as it was built using fewer advanced chips compared to leading AI models like OpenAI’s GPT-4.
The sudden rise of DeepSeek has sparked discussions about growing influence of China in AI and its ability to innovate despite US restrictions on advanced chip exports. The AI industry has long relied on high-performance chips, but the success of DeepSeek with a cost-effective approach is now challenging the belief.
DeepSeek functions similarly like the ChatGPT of OpenAI as it helps users with tasks like answering questions, writing and coding. However, it is trained to avoid politically sensitive topics. The efficiency of DeepSeek and low cost have drawn global attention. Some experts estimate that its training cost could be at just $6 million while GPT-4 spent more than $100 million.
The AI startup reportedly accumulated a stockpile of Nvidia’s A100 chips before they were banned for export to China. The company managed to build a powerful model at a fraction of the usual cost by pairing these chips with lower-end alternatives.
Meanwhile, DeepSeek faced large-scale cyberattacks on the day it became the most downloaded free app in the US. The platform also experienced website outages and this highlights the challenges of scaling up AI technology amid growing demand.