🐋

Liang Wenfeng

Original nameLiang Wenfeng

Chinese Entrepreneur, Founder of DeepSeek

Those Who Turned Papers Into What a Billion People Use · Industry & Product Builders

The DeepSeek Series of Large ModelsLow-Cost High-Efficiency Training MethodsThe Industrial Impact of Open-Source Large Models

Who they are

Liang Wenfeng (1985– ) is a Chinese entrepreneur who in his early years founded the quantitative investment firm High-Flyer, and in 2023 founded the AI company DeepSeek, putting the high-performance computing infrastructure and engineering-optimization experience accumulated in the quantitative-investment industry into the development and training of large language models. The model DeepSeek released in early 2025 achieved performance rivaling the world’s top closed-source models at a training cost far below the industry’s common expectation, and chose to release it with open weights, and after this news broke it caused a violent shock in global tech and capital markets, the share prices of chip companies such as NVIDIA falling sharply that day, and the market began to reexamine the previously widely assumed industry premise that "training a top large model must rely on piling up vast compute." Liang himself had rarely appeared in public before, and DeepSeek’s technical report and open-source strategy are widely seen as a representative case of the Chinese AI industry, in an external environment of restricted compute, turning to seek algorithmic-efficiency breakthroughs.

Primary sourcesDeepSeek’s official technical report (2025)Public financial reporting on Nvidia’s share price

Key stories

A Technical Report That Made the Market Recompute a Problem

In the technical report DeepSeek released in early 2025 it disclosed its training cost and engineering-optimization details in detail, claiming to have trained a system rivaling top closed-source large models with a fund investment far below international peers, and the day this news broke, the share prices of chip companies such as NVIDIA saw a historic single-day sharp fall, the global market beginning to reexamine the previously widely assumed industry premise that "the AI race must rely on piling up vast capital and compute," this shock making "efficiency," for the first time, as much a core dimension for examining the large-model race as "scale."

Relationships

Ally

🎓Kai-Fu Lee

Nemesis

🎮Jensen Huang 💬Sam Altman

Echoes today

Below are how modern works borrow or reinterpret this name or story — not the original material. The two differ, so keep them apart.

The "efficiency narrative" challenges the "scale narrative"DeepSeek’s shock thereafter made the efficiency view of "achieving the same effect with fewer resources" become, for the first time, as core a dimension for examining the large-model race as the scale view of "piling up more compute."

Appears in

The Generative-AI Explosion and the Industry ContestNvidia’s Decade-Long Bet and the Shock of DeepSeek

← Back to Figures