September 4, 2026 (InvestinChina.asia) – The Ministry of Industry and Information Technology (MIIT) has rolled out a sweeping three-year blueprint designed to turn China into a global hotbed of artificial-intelligence entrepreneurship. The AI Small-and-Medium Enterprise Startup Support Plan (2026–2028), issued on August 29 and made public this week, sets concrete targets that aim to double down on the country’s already aggressive push into next-generation intelligent technologies.
By 2028, the plan seeks to incubate more than 10,000 new technology-driven and innovative SMEs, elevate over 2,000 to the status of “little giants” under the government’s specialized and sophisticated SME program, and spawn a visible cohort of gazelle and unicorn companies in the AI sector. It also calls for building 10 high-standard technology business incubators, 10 national-level public service demonstration platforms for SMEs, and 10 national SME characteristic industrial clusters — all focused on artificial intelligence.
“The plan is about lowering the threshold and the cost of failure for AI entrepreneurs,” an MIIT spokesperson said in an accompanying interpretation, describing the desired outcome as a “tropical rainforest-style” innovation ecosystem where quality SMEs continuously emerge.
The policy attacks the problem from four angles — factor supply, entrepreneur cultivation, ecosystem enablement, and service guarantees — broken into 15 specific tasks.
Cheap Compute and Open Data
Top of the list is making computing power affordable. The plan leans on the National Computing Power Platform to pool and elastically allocate resources, delivering low-cost, efficient compute to startups. It deepens a dedicated initiative to extend inclusive computing power to SMEs, builds edge-computing facilities on demand, and upgrades dedicated matchmaking zones that connect compute supply with startup demand.
On the data front, the plan encourages industry leaders, platform companies, and trade associations to form data consortia with SMEs, building high-quality industry datasets in areas such as intelligent manufacturing, materials R&D, and biomedicine — then opening them to AI startups in an orderly fashion. Data service providers will be cultivated to offer consulting, governance, and labeling services, driving down the cost of data acquisition.
Scenario access is the third pillar. Through campaigns such as “Hundred Events, Ten-Thousand Enterprises” and the “In-Depth Tour” for AI empowering new industrialization, anchor enterprises will be guided to open up test-validation, demonstration, and large-scale commercial scenarios to startups, accelerating product iteration.
A New Breed of Entrepreneurs
In a notable nod to changing times, the plan explicitly embraces so-called “one-person companies” (OPCs) and super-individuals who leverage intelligent tools for agile entrepreneurship. It pushes to accelerate the cultivation of “AI-native” enterprises and supports open-source community core contributors and agent developers in finding commercialization paths.
Incubators themselves are told to evolve beyond rent-collecting office spaces. They are expected to embed smart-compute provision, industry data sharing, and scenario matchmaking into their core service offerings, and to experiment with flexible arrangements such as service fees and equity-based rent — creating risk-sharing, benefit-pooling communities with the startups they house.
Capital, Open Source, and Clusters
Financing gets a major boost. The plan puts the National SME Development Fund, the National AI Industry Investment Fund, and the National Integrated Circuit Industry Investment Fund to work as anchors, attracting social capital to set up early-stage VC and venture funds focused on AI. The goal is a “patient capital” system covering the seed, startup, and growth stages. A dedicated AI investment-and-financing section will be added to the gradient cultivation platform, with online roadshow capabilities to connect VCs and founders efficiently.
The plan also signals openness to novel support mechanisms — explicitly encouraging local governments and investment institutions to explore “compute equity,” “data equity,” and investment-incubation linkages, bundling capital with compute, data, and pilot resources.
Open-source ecosystem building is treated as a strategic priority. The national-level AI open-source community AtomGit will be strengthened, and AI startups are encouraged to contribute models, tools, and datasets. Quality open-source projects will receive targeted support from various investment funds, forming what the document calls a “virtuous-cycle open-source ecosystem.”
Cluster development rounds out the strategy. National high-tech industrial development zones are told to focus on new AI tracks and intensify SME cultivation. In leading regions such as the National AI Innovation and Application Pilot Zones, clusters will build shared infrastructure — including compute supply-demand matching platforms and inclusive compute empowerment centers — to make compute and data universally accessible.
The plan arrives as China’s 2026 government work report emphasizes improving employment and entrepreneurship measures adapted to AI technology development, and aligns with the broader “AI+” action agenda advanced by the State Council. For investors, founders, and multinational technology partners watching China’s AI landscape, the policy sends an unmistakable signal: the world’s second-largest economy is systematically removing the structural barriers that have historically kept small players from competing in frontier technology.