In the early hours of August 12, Lin Junyang announced on X the founding of Pragmatik Labs, based in Shanghai, pointing toward next-generation agents. Explaining the name, he wrote that he studied linguistics because a friend recommended it, then turned to computational linguistics and NLP, and that the name means returning to where things happen, and also pragmatism: we believe AI should aim for that. A few months after leaving Alibaba’s Qwen team, Lin has formally become an entrepreneur. Investors including Gaorong Ventures, HongShan, and Tencent appear on his financing list, with market reports placing the valuation at 2 billion US dollars.
Born in 1993, Lin holds a bachelor’s from China University of International Relations and a master’s from Peking University. He joined Alibaba DAMO Academy in 2019 and moved into the core Qwen team, leading development of the Qwen series into globally influential open-source models. On March 4 he announced his departure with me stepping down, bye my beloved qwen.
Lin’s startup lands at a notable moment: China’s technology industry is raising the position of scientists. More researchers are entering the cores of model teams, foundation model leaders sit close to the highest decision-making structures, and young researchers leaving big companies can quickly obtain capital on technical judgment. The knowledge scientists hold is becoming scarce industrial power again. These scientists have systematic training in computer science or AI, graduate research experience, or long-term frontier research. In industry, they stand between frontier research and company decisions, judging technical direction when no ready answer exists and turning research into engineering paths. The scarce value extends to resource allocation: compute, R&D budgets, and teams follow technical judgment.
China has seen this power shift before. When technology matures, markets, channels, and scale rise; when industry re-enters uncharted territory, those who find new answers move toward center stage. Similar changes are happening across companies: Baidu placed Wu Tian, head of foundation model R&D, directly in Robin Li’s reporting chain; Tencent brought in Shunyu Yao, a Tsinghua Yao Class undergraduate and Princeton PhD, to lead large models and AI infrastructure; ByteDance has Wu Yonghui, a Peking University undergraduate and HKUST PhD, leading Seed foundation model research. Among startups, Kimi founder Yang Zhilin holds a Tsinghua bachelor’s and Carnegie Mellon PhD, and MiniMax founder Yan Junjie earned his PhD from the Chinese Academy of Sciences Institute of Automation.
The pattern is visible across the industry: whether at large firms or startups, the technical arc of the AI era is putting researchers back at the center of power. China’s major model ecosystem increasingly resembles a scientist’s era, where frontier researchers who can see where the field is going also control how resources are deployed to get there.
