China is rolling out “computing power voucher” programs across numerous cities, including Beijing, Shanghai, Chengdu, Shenzhen, Shandong, Henan, and Ningbo. These vouchers subsidize AI computation costs for small and medium enterprises (SMEs), letting them rent data center time at significantly reduced rates, tells Tom’s Hardware. In Shanghai, for instance, local governments have allocated roughly CN¥600 million (about USD 84 million) to this program, offering subsidies covering up to 80% of AI rental fees. Shanghai also earmarked CN¥100 million for AI large-language model (LLM) training.
Chengdu, which piloted the initiative in 2023, has increased its commitment with a CN¥100 million fund aimed at research institutions. Meanwhile, Shandong province dedicated CN¥30 million and is preparing to invest an additional CN¥1 billion to bolster AI infrastructure. Beijing and Henan have launched their voucher schemes and are now accepting applications.
This program has roots in a December 2024 policy from China’s National Development and Reform Commission, promoting high-quality growth in sectors such as data labeling and AI development by lowering R&D barriers for smaller players.
China’s voucher initiative aligns with its broader strategy of building a national computing network. The country has seen a rapid expansion of data centers under its “Eastern Data, Western Computing” plan, creating capacity that often runs at just 20–30% utilization. These vouchers aim to match underused compute resources with demand from AI developers, enhancing utilization and bridging costs.
Why This Matters for Engineering and AI Innovation
- Lower barrier to entry: The vouchers let startups and SMEs access powerful compute resources without heavy upfront spending, fueling AI projects that might otherwise stall.
- Optimized infrastructure: By leveraging underused data centers, China improves efficiency, reducing idle capacity and waste.
- Boost to AI ecosystem: Subsidies spread compute access across cities, supporting geographically diverse innovation rather than concentrating it in tech hubs.
- Policy-driven agility: Through these local schemes, China demonstrates how regulatory outreach can directly influence technological growth.
- Scalability for future AI workloads: If successful, this model could scale nationwide, setting the foundation for AI-centric infrastructure on a massive scale.