Zhipu, MiniMax, and DeepSeek’s leaders of three Chinese AI companies have successively announced major plans around July. Their stock and valuation trends have diverged. As the head of “the world’s first publicly listed large-model company,” Zhipu Tang Jie announced the “Touch High” plan via an internal email after listing for half a year, asking the team to ease pressure from commercial monetization over the next two years and to focus resources on long-range missions, autonomous agents, and self-evolution foundational research.
Zhipu Tang Jie’s “Touch High” plan: ease commercial monetization pressure for two years, focus on three foundational research directions
As founder of the world’s first listed large language model company (02513-HK), Tang Jie announced the “Touch High (摸高)” plan via an internal email about half a year after listing. He made it clear that although the market has praised Zhipu’s market cap performance, Zhipu’s goal is not short-term financial windfalls, but a return to foundational research.
According to the internal email, the plan asks the team to ease pressure from commercial monetization over the next two years and to focus its core resources on the following three cutting-edge directions:
Long-range missions: strengthen the model’s planning and execution capabilities in complex, multi-phase tasks
Autonomous agents: develop an AI agent framework with autonomous decision-making capabilities
Self-evolution: research technical pathways for autonomous learning and iterative capability improvement in models
In his letter, Tang Jie said his methodology is “essentially counterintuitive and focused,” explicitly separating research from short-term market competition.
MiniMax Yan Junjie’s company-wide email: zero-salary statement, 4% equity incentive, and 1% open-source ecosystem support
During a period of turbulence in the company’s stock as it underwent a deep pullback, MiniMax founder Yan Junjie announced three specific measures through a company-wide email: until the company achieves its AGI goal, he will no longer receive any salary from the company; he will set aside 4% of his personal shares to incentivize the core team; and he will also use 1% of his personally held shares to support the development of the open-source ecosystem.
In his letter, Yan Junjie said that for core technical personnel who hold options, the founder’s public commitment is one of the key factors for stabilizing morale. All three measures above are voluntary adjustments to his personal holdings, not financial decisions made at the company level.
DeepSeek Liang Wenfeng’s funding pivot: valuation surges to $71 billion in a month
DeepSeek Liang Wenfeng completed a major strategic shift, moving away from its early stance of refusing to raise funds and not being bound by commercial timelines, to pursuing large-scale external fundraising. According to media reports, DeepSeek’s valuation surged to $71 billion within a month, and there have also been reports about its IPO plans; the specific listing schedule will be subject to official announcements.
Liang Wenfeng also led DeepSeek in building data centers and expanding its compute reserves. Liang Wenfeng’s public explanation of the strategic pivot shows that amid the dual pressures of compute constraints and global technology races, external fundraising has become a necessary path to expand compute. DeepSeek’s approach to completing model training with relatively lean R&D costs had already drawn widespread attention from international markets.
FAQ
What exactly does Zhipu’s “Touch High” plan require, and how long will it last?
According to Tang Jie’s internal email, the “Touch High (摸高)” plan asks the Zhipu team to ease pressure from commercial monetization over the next two years and to focus core resources on three foundational research directions: long-range missions, autonomous agents, and self-evolution. The plan was introduced about half a year after Zhipu (02513-HK) listed.
What are the trigger conditions for Yan Junjie’s zero-salary statement, and how many shares are involved?
According to Yan Junjie’s company-wide email, the trigger condition for the zero-salary statement is “before MiniMax achieves its AGI goal.” As for shares, Yan Junjie announced that he would set aside 4% of his personal shares to incentivize the core team, and also use 1% of his shares to support the open-source ecosystem; both are voluntary adjustments to his personal holdings.
How did DeepSeek’s $71 billion valuation form, and what is the IPO timeline?
According to media reports, DeepSeek’s valuation surged to $71 billion within a month, and there have been reports about IPO plans as well. DeepSeek’s approach to completing high-performance model training with relatively lean R&D costs is one of the main reasons it previously drew widespread attention from the international market; the specific IPO timeline will be subject to DeepSeek’s official announcement.
