September 3: Two Chinese districts have introduced new funding programmes aimed at helping companies access artificial intelligence computing resources and model services, offering different forms of financial support for AI development and digital applications.
Shenzhen’s Longgang District opened applications in late August for its 2026 AI computing-power support programme, while Beijing’s Tongzhou District introduced an implementation guide on September 3 covering support for the digital economy in the Beijing Municipal Administrative Centre.
Longgang’s programme operates under revised district measures adopted in June. It provides financial assistance to companies purchasing computing capacity from non-affiliated providers for large-model training, inference and generative AI applications.
Eligible companies are assessed through five funding tiers based on verified computing consumption and expert review, with consideration given to increases in production and computing capacity. Annual support ranges from up to CNY 4 million under the first tier to CNY 20 million under the fifth tier.
The programme is focused on reducing the cost of computing resources for companies developing and deploying AI models and applications. Companies must meet the programme’s eligibility and verification requirements to qualify for the subsidies.
Tongzhou has adopted a different approach. Under its new implementation guide, small and medium-sized enterprises purchasing AI model and computing services can receive model and computing vouchers worth up to CNY 500,000 per company annually.
The district is also providing financial support for demonstration projects involving technologies including artificial intelligence, cloud computing, the metaverse, big data, sixth-generation mobile technology and cybersecurity. Eligible projects can receive up to CNY 500,000, with annual support capped at CNY 2 million per company.
Additional support under the Tongzhou programme covers areas such as technology development, digital standards, innovation platforms and future-industry projects. Applications opened on September 3 and will remain open until October 8.
The two programmes illustrate different approaches to supporting AI adoption at the local-government level. Longgang places greater emphasis on direct support for companies purchasing computing capacity, with substantially higher funding limits for qualifying firms. Tongzhou combines smaller computing and model-service vouchers with broader support for digital-technology demonstration projects.
The measures also highlight an increasing focus on supporting demand for computing resources rather than solely investing in physical computing infrastructure. By helping companies meet the cost of model training, inference and digital applications, local governments can encourage businesses to make greater use of existing computing capacity.
However, the programmes are district-level initiatives rather than a nationwide AI computing subsidy programme. Their scale, eligibility conditions and funding limits differ, reflecting the priorities of the respective local governments.
The initiatives nevertheless provide an indication of how local authorities in China are using financial incentives to encourage companies to adopt AI and other digital technologies. As demand for computing resources grows alongside the development of large AI models and digital applications, such programmes could become one component of local strategies aimed at expanding AI adoption and supporting the digital economy.