Large Model Companies and AI Infrastructure Firms Recently Attract Capital


Academician Zhou Zhihua pointed out that there is a misunderstanding in the current research field of following large models to solve everything. He emphasized the need to optimize AI research planning. He suggested establishing interdisciplinary special zones to promote the deep integration of AI with other fields, avoid resource waste, and promote the healthy and diversified development of artificial intelligence.
On March 3rd, the official Weibo account of the open source project OpenClaw was launched, and its first post triggered interaction from multiple domestic large model manufacturers. The project has shown remarkable performance in the global AI field, is currently strong on GitHub's trending list, and has attracted attention during MWC2026.
Meta is secretly testing an AI shopping feature that leverages social data to challenge OpenAI and Google, offering personalized recommendations based on user profiles to transform the AI shopping experience, marking a shift in large model competition from content creation to e-commerce.....
As a pre-release version of V4, DeepSeek V4 Lite has attracted attention with 200 billion parameters and a context length of up to 1 million tokens. After continuous upgrades, its performance is comparable to top closed-source models, showing outstanding results in various benchmark tests and demonstrating strong competitiveness.
The domestic large model sector has seen a significant integration. Mianbi Intelligence has introduced a telecommunications giant and an industry fund, completing in-depth equity binding. This move not only brings financial support but also accelerates the commercialization of domestic large models in the fields of public data and smart hardware, marking the deep integration of "national team" and Tsinghua University-related technical forces.