Revealing the secret formula behind DeepSeek's huge success

The Chinese AI company made a splash when it launched its DeepSeek assistant, a formidable rival to ChatGPT, and quickly became a global phenomenon. What's more amazing is that DeepSeek achieved this feat while under a US embargo, cutting off access to the latest GPUs for AI development.

They did this by developing their own software tools to train AI at a significantly lower cost than their competitors, a success that has wiped $1 trillion off the US stock market amid concerns that hardware will no longer be the most important factor in AI development.

Not stopping there, DeepSeek is rushing to complete the DeepSeek R2 upgrade, scheduled for release next May. Although still under embargo, this new R2 model will make the most of the existing GPU inventory and software optimizations.

But the real secret behind DeepSeek's success lies not just in software or hardware. According to Reuters, the company has a unique work environment that is different from the "996" culture (working from 9 a.m. to 9 p.m., six days a week) common at Chinese tech companies.

DeepSeek founder Liang Wenfeng built the company as a research lab, encouraging creativity and allowing employees to work eight hours a day. He also worked directly with young engineers, giving them ownership of research projects and paying them well.

In addition, DeepSeek has received strong support from the Chinese government. Their AI applications are widely used in government agencies and large technology companies such as Baidu, Lenovo and Tencent.

But the company also faces major challenges. Concerns about data privacy and censorship have led to the app being banned in countries like Italy and South Korea. The US is also considering a broader ban.

Still, the DeepSeek R2 launch in May is sure to draw a lot of attention. Will the R2 be able to continue to create the same shock as its predecessor? The answer will soon be revealed.

Samuel Daniel

Update 03 March 2025