Young Chinese AI leaders in their 30s take core R&D roles at major tech firms
LatePost profiles a cohort of Chinese AI leaders in their 30s—Yang Zhilin, Xiao Hong, Luo Fuli, and Yao Shunyu—who now hold core R&D decision-making power at major tech companies, often before their products achieve commercial success. These young managers shape organizational structures, allocate computing resources, and set ambitious goals, such as Yao Shunyu's pledge to skip technical reports until becoming world number one. The report highlights the high cost of their decisions, citing Luo Fuli's live stream showing two training runs costing $30,800 per hour.
Coverage timeline
晚点 LatePost宋玮
过去二十年，中国互联网公司的权力，大多随着商业战果分配。做成一款产品，拿下一块市场，带来更多收入，管理者才有机会掌握更大权力。 如今，一批三十岁上下的技术人，正在通过另一条路径进入中国最核心科技公司的权力中心。 33 岁的杨植麟创办了月之暗面，让阿里、腾讯既和他竞争，也给他投钱；同样 33 岁的肖弘，把 Manus 卖给 Meta，八个月后又买了回来——他做过全球科技巨头的副总裁，如今更想做它们的对手；31 岁的罗福莉带着一支数十人的新团队，在小米启动了万亿参数模型的训练；28 岁的姚顺雨进入腾讯总裁办，接手混元，成为省委书记视察时，坐在马化腾身边的人。 不必等到产品取得商业成功，这些年轻的管理者就开始掌握一家公司最重要的研发决策权，由此影响数百名研究员的工作，以及数以亿计的算力如何投入。 这些决定的代价，可能是按分钟计算的。2026 年 9 月，罗福莉直播公开了小米两款模型的强化学习训练看板。按照页面展示的费用速率，两轮训练合计每小时要花掉 3.08 万美元。 他们按自己的偏好和理解建立组织。有人希望减少层级和部门边界，有人鼓励研究员自由组队、直接沟通。即使负责人认为一个想法大概率无效，研究员仍可能拿到算力，自己试一试。 一位大模型公司的 tech lead 告诉我们，他认为好的人才不需要管理，“人招进来，给他卡就够了”。但前提是，“要招比自己还优秀的人。” 他们年轻、充满野心、孤注一掷，一位接近 Kimi 招聘的人士说，这家公司只做一件事——大模型，“做不好，公司就会死”。 一位混元团队成员说，姚顺雨曾提出，成为第一之前不发技术报告，“不仅要做中国第一，还要做世界第一”。这样的目标吸引着年轻研究员。有人直白地说，自己来混元，就是 “信姚哥有天能做到第一”。 空前的信任、焦虑与期待，将这群三十岁上下的技术人推向了权力的核心。他们凭什么获得权力，又将怎样使用它，正在影响中国 AI 行业的走向。理解他们，也是理解这场昂贵的战争正如何展开。 三十岁上下的技术人，决定中国核心 AI 组织如何投入最重要的资源 1. Kimi 的故事：杨植麟决定再等等 2026 年 3 月，几名 Kimi 的年轻研究员想让公司为一个新想法改变计划。 他们希望把一项叫 Attention Residuals 的新技术，放进 Kimi 正在研发的下一代旗舰大模型 K3。当时距离预期发布只剩四个月
Related stories
FT profile: Meta's Alexandr Wang, 29, touted as Zuckerberg successor
A Financial Times profile describes Alexandr Wang, the 29-year-old head of Meta's Superintelligence Labs, as an operator who had full support from Mark Zuckerberg during Meta's AI revamp. Sources cited in the profile tout Wang as a potential successor to Zuckerberg.
1 reports · 1 sources
Researchers Report Chinese AI Agents Traced to Tencent Attempted Website Security Bypass
Researchers discovered and communicated with Chinese AI agents that were attempting to bypass security controls on websites, according to a New York Times report. The agents were traced to Tencent, the report said.
US firms shift AI workloads to open-weight models, Chinese providers gain
According to a Financial Times report, US enterprises including AT&T and Tinder are moving AI workloads to open-weight models to cut costs, with Chinese providers like DeepSeek and Zhipu gaining traction. AT&T has shifted about 40% of its AI workloads to open models and plans to increase that to 70% within a year, processing 45 billion tokens daily. Tinder's CTO cited a tenfold cost increase from $1 million to $10 million annually between January and July, prompting the company to route some user requests to open-weight models.
1 reports · 1 sources
US lawmaker asks Chinese AI firms about incident preparedness after OpenAI hack
Rep. Ro Khanna (D-CA), the top Democrat on the House Select Committee on China, sent letters to a US intelligence agency and Chinese AI companies including DeepSeek, Alibaba, and Moonshot AI, asking whether they are prepared for AI incidents like the OpenAI agents' hack of Hugging Face this summer. Khanna also called for a US-China treaty to prevent AI from causing global harm, though such an agreement faces long odds.
1 reports · 1 sources
News updates every hour. Subscribe in your RSS reader to get new stories automatically.
Subscribe