JAKARTA – Behind the meteoric rise of the Chinese startup DeepSeek lies a team of young developers working on an AI model set to rival OpenAI’s ChatGPT from the United States (US).

DeepSeek’s founder, Liang Wenfeng, began his journey in the AI field through Fire-Flyer, a deep learning research project under High-Flyer, a hedge fund established in 2025 with an initial capital of USD 15 billion.

A few years into its operations, High-Flyer had amassed a significant supply of graphic processing units (GPUs), initially used for financial data analysis by Fire-Flyer.

Wenfeng later leveraged these financial resources and equipment to build DeepSeek, culminating in the release of its latest model (R1) on 20 January. This new model has shaken the US stock market, being priced lower than ChatGPT and offering open-source code to the public.

In an interview with China’s local media outlet 36Kr, Wenfeng stated that DeepSeek's development was driven purely by a passion for knowledge rather than profit. “Commercially, it’s not valuable. The return on investment (ROI) for scientific research is very low,” Wenfeng said.

For this reason, Wenfeng involved young people who share the same vision in developing DeepSeek. “Our core technical team is mostly made up of people who have just graduated in the last 1–2 years,” Wenfeng explained.

He believes this strategy fosters a culture of collaboration, where team members gain free access to enough resources to work on unconventional projects. This approach contrasts sharply with other players in China’s tech industry, who compete fiercely to acquire the most superior resources.

Wenfeng even emphasised that he prefers to place creative young people, rather than those with experience, at the forefront of DeepSeek's development. For him, creativity and the drive to grow are more valuable than work experience. (KR/ZH)