张小珺·商业访谈录
The most solid long-form AI and business interviews in Chinese
2:34:18The AI Leaders Are Only Refineries, Not the Native-App Winners
AI has entered the agent stage: the leaders of stage one are only refineries, not the big winners of native applications; new platforms do not come from old platforms; organizations have to shift from jobs to tasks; and in the end what people compete on is the creativity to make something out of nothing.
2:04:19K3 breaks no new ground, but pushes activated parameters to the 100B class
Sun Yutao walks through the Kimi K3 technical report and argues it has no single breakthrough: it fuses techniques that already existed — linear attention, hybrid attention, dynamic load balancing — and gets activated parameters up to the 100B class. His judgment: large models will see no more Transformer-style invention, only refinement.
3:36:12NVIDIA's VP of Research: The Endgame for Models Is Ecosystem, Not Beating Rivals
Ming-Yu Liu: model capabilities will eventually converge, and NVIDIA's open-source world model is not competition but groundwork; physical AI's GPT moment has not arrived, but the conditions are converging.
1:40:35The Next Breakthrough in AI Research: Build a Physics of AI First, Then Let It Discover Architectures Itself
Liu Ziming argues that real AFAI is not writing papers or rewriting training pipelines, but using the methods of physics to structure research first, then training a meta model to predict training curves, so that AI eventually discovers the next generation of architectures on its own.
3:00:26Open models will win, and vLLM is the Linux of AI inference
vLLM is already the most active open-source project on GitHub, but Kaichao You's (游凯超) real claim is this: model capability cannot stay hidden for long, open models will win in the end, and the inference engine is the next Linux — where the moat is fast iteration, not the model itself.
1:52:54Embodied AI hasn't reached its GPT-1 moment; the block is data that won't scale
Shen Yujun (沈宇军) of Ant Lingbo (蚂蚁灵波) argues embodied AI has not reached its GPT-1 moment, and the bottleneck is that data has not scaled up: internet data is two orders of magnitude larger than real-robot data, and an ideal pretraining run would start at a million hours. To get there the team rebuilt visual pretraining from scratch and insists on starting from real sensors.