On the banks of West Lake in Hangzhou, a new kind of technology space is quietly trying to solve a familiar problem: how to turn model research and open‑source activity into paying industrial use. The Moda Community Developer Center — the first brick‑and‑mortar outpost of China’s largest open‑source AI model community — opened in November 2025 with more than 10,000 square metres of space intended to host developers, startups and established companies side by side.
The centre, built through a public‑private partnership led by Alibaba Cloud Valley and local government resources, integrates shared offices, AI experience exhibits, a demo hall and technical support for model downloads, API access and localized deployments. Rents are deliberately low — some co‑working desks are priced at about 300 yuan a month — and the centre offers end‑to‑end services from scene matchmaking to company registration and market exposure for founders.
That practical orientation matters because Zhejiang’s provincial leadership has made industrialising AI a policy priority. The 2026 Zhejiang government work report explicitly pledged support for Moda to become an international‑class open‑source community. Zhejiang’s own AI sector produced roughly 680 billion yuan in core industry revenue in 2025, growing about 20% year‑on‑year, and the province aims to sustain similar growth in 2026.
The centre’s operators have been deliberate about knitting together supply and demand. They host industry contests, salons and forums — 18 events to date with more than 5,000 participants — and maintain a “scenario opportunity” database that aggregates 292 government and state‑owned enterprise opportunities alongside 362 firm‑level capability listings. Open‑source learning communities such as Datawhale have moved in, arguing that face‑to‑face interaction will accelerate the transfer of ideas into deployable products.
Startups already resident in the space tell a similar story. Entrepreneurs say the centre helps them answer concrete questions that keep traditional firms from adopting AI: which business process to automate, which base model to use and how to manage data security when moving analytics workloads to AI. The centre supplies localized tutorials and tool chains intended to lower the technical threshold for on‑premise deployment — an increasingly salient capability as Chinese firms weigh data security and sovereignty concerns.
The initiative is neither purely commercial nor purely civic. By combining municipal scene inventories and government project lists with Alibaba’s cloud capabilities and a prominent open‑source community, Zhejiang is building an ecosystem designed to fast‑track industrial pilots into scaled deployments. That convergence of ruling‑party priorities, corporate infrastructure and grassroots developer networks is a template Beijing has encouraged across other tech clusters, but Zhejiang’s focus on open‑source communities gives its approach a distinctive flavour.
For international observers, the centre is a reminder that China’s AI strategy is as much about industrial adoption and ecosystem building as it is about frontier model research. Efforts like Moda’s lower the barriers to experimentation for small teams and medium‑sized firms while generating a local feedback loop of talent, demos and paying customers. They also reinforce the role of dominant cloud providers in shaping which technical stacks and deployment models become standard across the Chinese market.
The immediate promise is heightened innovation velocity; the longer‑term implications include regional lock‑in of preferred vendors, tighter alignment between firms and municipal priorities, and a more elaborate pathway for state‑backed tech to reach industry. How open the community’s outputs remain, and whether the model attracts genuine cross‑border collaboration or accelerates a distinct, China‑centric AI stack, will determine whether this experiment proves globally influential or mainly locally consequential.