Shanghai AI Hub Nurtures Next-Gen Model Startups and Brain-Inspired Chips

Model Speed Space, an artificial intelligence incubator in Shanghai's Xuhui District, is driving commercial deployment for early-stage ventures spanning vision chips, multimodal generation, and robotics. Established as a focal point for foundation model development, the community recently saw Shanghai Xijian Technology tape out a low-power brain-inspired vision chip and Daxiao Robotics deploy world model solutions for automated retail. Backed by corporate challenge programs with giants like Huawei and Alibaba Computing, the hub reflects Chinese policy efforts to channel young engineering talent into concrete industrial applications.

Model Speed Space illustrates how municipal planners and state-backed tech ecosystems lower barriers for generative AI startups. By centralizing computing infrastructure, specialized datasets, and matchmaking with heavy industries—such as shipbuilding, transportation, and infrastructure—the hub aims to solve the commercialization bottleneck that frequently traps early-stage foundation model companies in proof-of-concept stages.

Shanghai's flagship foundation model incubator, Model Speed Space, is accelerating product deployment across robotics, enterprise agents, and specialized silicon, according to state media reports detailing the hub's recent progress. Located in Shanghai's Xuhui District, the innovation community serves as an end-to-end ecosystem integrating compute resources, foundational models, and hardware developers. Among recent technical advances, Shanghai Xijian Technology completed the tape-out of its second-generation brain-inspired vision chip designed for robotic systems. The chip achieves a resolution of 5 megapixels while reducing power consumption by roughly 80 percent compared to traditional vision sensors. The progress follows high-level policy attention directed at the facility. In April 2025, Chinese President Xi Jinping visited Model Speed Space, emphasizing that artificial intelligence represents an enterprise tailored for the younger generation and urging founders to align personal innovation with broader national modernization goals. To bridge the gap between algorithmic research and real-world deployment, the hub launched an open challenge initiative in May. Tech incumbents and research institutes—including Huawei, Alibaba Computing, and the Shanghai Artificial Intelligence Laboratory—posted 32 real-world engineering bottlenecks. Multi-disciplinary teams across China were selected to tackle these tasks, with high-performing projects advancing directly toward commercialization and venture incubation. Hub administrators have also focused on scenario matchmaking, guiding startups into industrial environments that are typically inaccessible to young teams, including maritime construction, transit hubs, and municipal infrastructure. Zhang Yun, party secretary of the innovation community, noted that high turnover in office nameplates reflects companies scaling out into larger standalone facilities as their commercial footprint expands. Tenant companies are increasingly targeting specialized operational niches. Shanghai Liblib Technology, founded by 34-year-old entrepreneur Chen Mian, has expanded its team to more than 200 staff while developing an autonomous multimodal video generation agent that breaks down scripts, characters, and shot transitions. In autonomous systems, Daxiao Robotics unveiled four world model solutions targeting automated retail environments during the 2026 World Artificial Intelligence Conference. Other resident enterprises include Xiangfeng Technology, which adapts meteorological models for agriculture and energy, and Qidian Zhiwei, which focuses on AI-assisted pharmaceutical development.