Sources: China is drafting plans to spend ~$295B over the next five years on building AI data centers, sourcing 80%+ of tech from local suppliers like Huawei
Context & Ripple Effects
This is the latest step in a long-running industrial-policy arc: China’s 2017 AI plan set domestic capability targets, while later reporting showed major buyers such as ByteDance increasing purchases from Chinese chip suppliers.
The policy environment has become more prescriptive. Guidance reported in late 2025 limited state-supported new data-center projects to domestic AI chips, and the 2026 five-year blueprint broadened the push through an “AI+” program.
First-order effects
- Huawei and other local infrastructure suppliers would gain a substantially larger, policy-backed pool of demand for AI chips, servers, networking, and data-center equipment.
- Operators building covered capacity would need to design procurement around local technology, rather than treating foreign components as the default option.
Second-order effects
- Large committed demand would give domestic suppliers more deployment feedback and production scale, strengthening their position with Chinese cloud, internet, and enterprise customers.
- Foreign AI-infrastructure vendors would face a narrower addressable market in China, while domestic buyers may accept more supplier concentration as a trade-off for policy compliance and supply-chain alignment.
Third-order effects
- If implemented at the indicated scale, the program would deepen the separation of AI-compute supply chains: China would be building an integrated domestic demand base alongside its own chip and data-center stack.
- The durable question is whether mandated local procurement translates into competitive performance beyond protected projects; the policy can create scale and adoption, but does not by itself establish technological parity.
The trend: AI infrastructure is becoming a strategic industrial-policy asset, with governments using data-center investment and procurement rules to localize the compute stack.