Hangzhou is pursuing the title of China's top city for artificial intelligence innovation, aiming to establish a pioneering pilot base for AI applications with unconventional mechanisms and policies.

This "pilot base" concept is critical because it tackles a specific bottleneck in AI development. While fundamental research moves from 0 to 1, piloting (or "中试") bridges the gap from 1 to 100. This stage is notoriously difficult, requiring large investments, long timelines, and facing high failure rates, often deterring capital.

For AI, the piloting challenge is particularly acute. Take embodied intelligence as an example. Transitioning robots from labs to factories, warehouses, hospitals, and homes involves more than algorithms. It requires a complex mix of chips, sensors, components, data, computing power, models, testing certifications, and real-world application scenarios. Companies can rarely shoulder this entire burden alone. The rapid pace of technological iteration amplifies this. Large models and embodied intelligence technologies evolve every few months. If a pilot platform follows traditional, slow bureaucratic processes for approval, construction, and procurement, it could become obsolete before it's even operational.

To accelerate this, Hangzhou is adopting a mixed-ownership model for the pilot base. Two state-owned entities, Hangzhou High-Tech Innovation Group and Hangzhou Data Group, will hold a controlling stake, with private companies like Unitree Robotics and Transfar Zhilian taking minority shares. The state will provide foundational investments like public training grounds, data platforms, computing power, and testing. Unitree will contribute robot technology and industrial resources, while Transfar will open up real-world logistics and manufacturing scenarios. This creates a structure of "state building the stage, enterprises performing, and market-driven pricing."

Research organization is also changing. Hangzhou is implementing a "four questions, one evaluation" system: companies pose real needs, government coordinates resources, universities and the pilot platform tackle problems, results are tested on the production line and in the field, and finally, the market provides the ultimate evaluation. The pilot base's core task is to offer scenario validation services to accelerate the scaling of solutions. This "unconventional" approach reflects Hangzhou's underlying philosophy: the government doesn't directly manage technological innovation but builds public infrastructure, organizes application scenarios, and shares the costs of trial and error, letting the market decide which technical paths succeed.

Compared to other industries, AI demands more flexible resource allocation and institutional experimentation, hence the need for "unconventional" measures. Many cities are already offering familiar tools like computing power subsidies, industrial funds, open challenges, application scenario access, and government procurement. However, a risk emerges: as more cities adopt these "unconventional" methods, they may become the new "conventional," leading to a new wave of industrial competition. This could result in redundant platforms, funds chasing the same projects, companies negotiating subsidies between cities, and applications remaining mere demonstrations without generating stable commercial orders.

Hangzhou, home to the "Six Little Dragons" of AI, is undoubtedly in the first tier. In the May 2026 Forbes China Top 50 AI Tech Companies list, 7 Hangzhou companies were selected, ranking third nationally behind only Beijing and Shanghai. This is a sharp increase from just two companies two years prior. In the large model track, Hangzhou ranked first among all cities by number of selected companies. Experts suggest that becoming China's leading AI innovation city is a "leapable goal" for Hangzhou. The pilot base could be its first springboard.

Currently, Hangzhou is the only city where a mixed-ownership company with both state and private capital has taken on a national-level AI platform, making its "unconventional" approach genuine. But the long-term challenge remains: can this platform truly reduce individual company innovation costs and solve industry pain points? More innovative mechanisms are needed.

Data underscores the opportunity. In the first half of this year, AI product exports from Yangtze River Delta cities grew remarkably. Hangzhou's exports of intelligent biomimetic robots accounted for 56.9% of the national total. As a key hub for robot production and export, the city hosts over 700 upstream and downstream companies involved in robot R&D, component manufacturing, software algorithms, and application scenarios. However, previous research noted that while Hangzhou's robotics industry is strong in hardware, it still lags in intelligence. The key question remains: how can robots become smarter and move from labs to production lines and finished products faster through effective "piloting"? This challenge still awaits new solutions.