The headline number is $295 billion, but the actual scope is wider. The goal is to connect scattered data facilities into a cohesive national network by 2028 — think of it less as “building more data centers” and more as China wiring itself for AI the way the US built the interstate highway system for cars.

Key agencies including the National Development and Reform Commission (NDRC) are drafting the blueprint. State-owned giants China Mobile and China Telecom will operate most of the infrastructure. And critically, the plan is part of Beijing’s broader “Six Networks” program — a national initiative covering water, electricity, computing, and other essential infrastructure. AI isn’t being treated as a tech sector story. It’s being treated as national infrastructure, on par with power grids and roads.

The most significant detail buried in this announcement is the chip mandate: at least 80% of the technology must come from domestic suppliers. That’s a direct shot at Nvidia, which has dominated AI hardware globally — and which has spent the past two years navigating US export restrictions that limited what it could sell to Chinese customers.

Nine categories of domestically developed AI chips from Huawei, Alibaba, Shanghai Biren Technology, and Moore Threads have already passed China’s security reviews. That’s a prerequisite for deployment across government and security-sensitive sectors — and with this plan, those sectors are about to get a lot bigger. Washington did recently allow Nvidia to sell its previous-generation H200 chips to China, which is one generation behind the current Blackwell architecture. Whether that concession matters in a world where China is actively building around it is now a real question.

At first glance, $295 billion over five years sounds enormous. And it is — it’s one of the largest state-directed technology investments ever announced by any government. But the comparison with US spending puts it in perspective: American hyperscalers — Meta, Microsoft, Google, and Amazon — have collectively earmarked over $725 billion for AI in 2026 alone.

That’s not a fair apples-to-apples comparison (government-directed spending vs. private corporate capex), but it does illustrate the scale gap. China’s advantage is coordination — the ability to direct state resources toward a single strategic goal in a way private markets can’t replicate. The US advantage is speed and private-sector innovation. What’s genuinely unclear is which approach produces better AI infrastructure faster. We’re essentially running a real-world experiment on two different models of building compute.

Here’s the scenario that keeps people in Washington and Brussels up at night. If China successfully builds a sovereign AI infrastructure — domestic chips, domestic data centers, domestic models — and connects it to its Belt and Road partners in Southeast Asia, Africa, and the Middle East, you end up with two separate AI ecosystems that don’t talk to each other.

One runs on Nvidia, AWS, and Google. The other runs on Huawei, China Mobile, and domestic equivalents. Developers, companies, and governments in the middle will have to choose sides. That’s not just a tech story — it’s a geopolitical realignment. The $295 billion plan doesn’t guarantee China succeeds. Huawei’s chips are still behind Nvidia’s best hardware in raw performance. But the direction of travel is unmistakable, and this investment is designed to close that gap over five years.