China's $295B AI Plan Backs Huawei, Locks Out Nvidia
Key insights
- China's NDRC is drafting a plan to spend 2 trillion yuan ($295 billion) on nationwide interconnected AI data centers over five years.
- State firms China Mobile and China Telecom are designated to operate the majority of the infrastructure under the plan.
- A mandatory 80% domestic technology sourcing rule targets Huawei as primary supplier while effectively excluding Nvidia and AMD.
Why this matters
A $295 billion state-directed mandate with an 80% domestic sourcing rule creates a captive procurement market for Huawei chips at a scale that could accelerate their technical development far faster than organic demand alone. For AI practitioners and founders, this formalizes the bifurcation of the global AI hardware market into distinct US-aligned and China-aligned supply chains, forcing infrastructure and tooling decisions that were previously optional. Private-sector spending from Alibaba and Tencent sits on top of this figure, meaning aggregate Chinese AI compute investment substantially exceeds the headline number and will reshape global model training capacity benchmarks.
Summary
China is preparing a 2 trillion yuan ($295 billion) plan to build AI data centers nationwide over five years, with China Mobile and China Telecom as primary operators.
The National Development and Reform Commission blueprint requires at least 80% of technology, including AI chips, from domestic suppliers like Huawei, effectively excluding Nvidia and AMD.
Essentially: (China Mobile, China Telecom, Huawei) get the contracts; (Nvidia, AMD) get locked out.
- The $295 billion covers state-directed spending only; private commitments from Alibaba and Tencent make China's real total larger.
- The 80% domestic sourcing rule makes chip localization a structural contract requirement, not just a policy preference.
State capital funds the backbone; domestic chip localization determines who builds it.
Potential risks and opportunities
Risks
- Huawei faces a delivery bottleneck if it cannot manufacture AI chips at the volume required by China Mobile and China Telecom across a five-year buildout timeline
- Western AI developers and hyperscalers lose access to workloads routed through the state-operated hubs, deepening the US-China compute divide and making cross-border model benchmarking harder
- If Huawei chips underperform Nvidia equivalents at scale, the 80% mandate could produce suboptimal infrastructure that costs China competitive ground in frontier model development
Opportunities
- Huawei's chip division and domestic suppliers gain a guaranteed multi-year procurement pipeline, funding R&D at a pace that could narrow the gap with Nvidia faster than market forces alone would allow
- Chinese AI startups and model developers gain access to heavily subsidized domestic compute, reducing infrastructure costs and enabling competition with US hyperscalers on training budgets
- Non-US chip equipment and materials suppliers outside the scope of US export controls may find expanded procurement opportunities within the buildout supply chain
What we don't know yet
- Whether the 80% domestic sourcing threshold applies at program launch or phases in gradually over the five-year timeline
- Technical performance gap between Huawei's current AI chips and Nvidia equivalents, and whether the blueprint includes any minimum performance benchmarks or capability floors
- How private-sector operators like Alibaba and Tencent will coordinate with state-operated hubs, and whether they face the same 80% domestic sourcing requirements
Originally reported by bloomberg.com
Read the original article →Original headline: China Plans $295 Billion Nationwide AI Data Center Investment — NDRC Blueprint Mandates 80% Huawei Chip Reliance, Squeezing Out Nvidia and AMD