DeepSeek's server burned down due to overload, forced to tighten belt

The popular AI startup DeepSeek is facing a difficult problem with server performance.

After causing a stir with the launch of the R1 AI model on January 20, DeepSeek continues to "make waves" in the market, attracting a huge number of users. However, this popularity also brings many challenges for the young startup.

DeepSeek's server burned down due to overload, forced to tighten belt Picture 1 DeepSeek's server burned down due to overload, forced to tighten belt Picture 1

DeepSeek restricts users due to overload.

DeepSeek recently announced that it was temporarily restricting access to its application programming interface (API) service due to 'server resource constraints.' The company said the move was to ensure stable operations for existing users while also working to expand its infrastructure to meet growing demand.

In addition to restricting API access, DeepSeek also announced an increase in chatbot service prices starting February 8. Specifically, the price for each million input tokens will increase to $0.27 and $1.10 for each million output tokens.

DeepSeek's meteoric rise has sent shockwaves through the US stock market. As soon as DeepSeek's chatbot reached the top spot on the App Store, major tech stocks, including the giant NVIDIA, lost trillions of dollars in value.

Faced with competitive pressure from DeepSeek, OpenAI has also responded by launching the o3-mini model and the Deep Research feature for ChatGPT.

DeepSeek is a clear demonstration of the strong development wave of AI. However, scaling the system to meet market demand is a difficult problem that this startup needs to solve.