New Lab Organization

New Lab Organization is the hybrid research-company form described in 149. 亲历中美 New Labs 资本狂潮，和清华刘子鸣聊：AI for AI、机制可解释性和 Max Tegmark. The source says 2026 brought a wave of young, elite-research-background AI teams in China and abroad that can raise at large valuations before ordinary product proof, especially around AI For Science, World Models, and AI For AI.

Liu Ziming defines the form by its temporary ambiguity. A New Lab begins closer to a research institute, giving itself six to twelve months for R&D, demos, and product-shape discovery. If it “groks” the route, it can then operate more like a company. The source argues this may be necessary when a direction is too risky for large companies and too slow for traditional academia.

Key Claims

• New Labs target pain points that the current dominant AI paradigm does not solve well.
• Their early advantage is belief, risk appetite, speed, and dense young research talent rather than mature product distribution.
• The form relies on venture capital accepting research uncertainty for a limited period.
• Auto Research and world models are especially suitable because they require research before easy engineering scale-up.
• The form can create a tension between scientific exploration and product/commercialization pressure after the first research window.

Connections

• Yuanhuan Intelligence, Liu Ziming, Tsinghua University, and Shanghai Qi Zhi Institute — source case.
• AI For AI, Auto Research, AI For Science, and World Models — hot directions named in the source.
• Academic AI Research Role and AI Organization Design — adjacent organizational frames.
• Research Taste and Training Compute Allocation — operating constraints for research-heavy teams.