The Origins of DeepSeek
DeepSeek's Liang Wenfeng built a competitive AI lab by stockpiling Nvidia A100s and reworking training to reduce GPU reliance after US export controls.
“Look, the way this works is we're going to tell you it's totally hopeless to compete with us on training foundation models. You shouldn't try, and it's your job to like try anyway. And I believe both of those things.”
A retrospective on DeepSeek's origins recounts how hedge-fund founder Liang Wenfeng pivoted High-Flyer toward AGI, amassed thousands of Nvidia A100 GPUs, and re-engineered training to cut GPU dependence after US chip-export limits. It underscores how a non-incumbent overcame compute constraints to challenge foundation-model giants, a defining signal about the viability of efficient, lower-cost model training.