Bold Claim: Liang Wenfeng, DeepSeek’s founder and CEO, has earned a place among the 10 people who shaped science in 2025, spotlighted by Nature for driving a new wave of AI innovation. At 40, the Hangzhou-based entrepreneur is praised as a “Chinese finance whizz” whose breakthrough AI models have stunned the world.
Nature’s profile emphasizes the January breakthrough of DeepSeek-R1, a reasoning model that challenged the notion that the United States was far ahead in AI—suggesting the gap may be smaller than widely believed. DeepSeek, a spin-off from Liang’s High-Flyer Quantitative Fund, disrupted the US stock market early in the year. On January 27, a sweeping sell-off erased nearly US$1 trillion in tech equities, including about US$600 billion from Nvidia, highlighting the move’s market-wide impact.
The company’s December releases, DeepSeek’s R1 and V3, delivered performance comparable to OpenAI’s offerings but at a fraction of the training cost. This undermined some of the valuation assumptions surrounding US semiconductor and AI players and raised questions about who actually bears the highest costs in AI development. For instance, Meta Platforms’ Llama 3 405B model reportedly required more than ten times the training expenditure.
DeepSeek’s open-source models—which researchers can download and adapt for free—have been a boon to experimentation and field-specific customization. Nature notes that these freely accessible models have spurred other firms in China and the US to release their own open-source variants.
Adina Yakefu of Hugging Face, quoted in Nature, calls DeepSeek “hugely influential,” underscoring the company’s broad impact on the AI landscape and on conversations around open, accessible AI systems.
