The largest single-nation AI infrastructure commitment in history just landed — and it didn't come from Silicon Valley. China's National Development and Reform Commission is finalising a five-year, 2-trillion-yuan ($295 billion) plan to stitch thousands of scattered data centers across the country into one unified national AI computing grid. The deadline for connectivity: 2028. The catch for US chipmakers: they're not in the room.
What Beijing Is Actually Building
This isn't a vague ambition document. The NDRC blueprint calls for a physical network of interconnected computing hubs, designed to function as a national AI backbone the way power grids or highway systems function as national infrastructure. State-owned firms China Mobile Ltd. and China Telecom Corp. are designated to operate the majority of the data centers and ensure they stay connected. Think of it less as a collection of company projects and more as a deliberate sovereign computing stack.
The 80% Domestic Chip Mandate — and What It Kills
Here's the headline that matters for global AI competition: the plan requires at least 80% of the technology powering these data centers — specifically including AI accelerator chips — to come from Chinese domestic suppliers. That means Huawei Technologies' Ascend chip series is positioned to become the backbone of one of the world's largest AI computing infrastructures. Nvidia, AMD, and their export-restricted products are effectively shut out at policy level, not just by sanctions. Analysts at Tom's Hardware note the irony: the same US export controls designed to slow China's AI progress may have accelerated Beijing's urgency to build a fully sovereign alternative.
The $740 Billion Hidden Number
The headline $295 billion covers computing infrastructure. But when power-grid integration is included — the transmission lines, energy systems, and power facilities needed to actually run thousands of large-scale data centers — the total projected investment could reach at least 5 trillion yuan, approximately $740 billion. That context transforms the story from "large tech spending plan" to "generational infrastructure commitment." For reference, the entire US interstate highway system cost roughly $500 billion in today's dollars. China is planning to spend more than that on AI infrastructure alone.
Can Chinese Chips Actually Carry This Load?
This is where the plan meets its hardest constraint. SMIC, China's most advanced domestic chip foundry, cannot yet produce AI accelerators at the performance density of TSMC-fabricated Nvidia H100s or Blackwells. Huawei's Ascend 910C is competitive for many workloads but still trails on raw FLOPS for frontier model training. The 2028 timeline is ambitious. Some analysts believe the computing targets are achievable only if domestic chip output scales significantly faster than current projections — and that's a manufacturing challenge, not just a policy one. Beijing is betting that necessity, capital, and time will close the gap.
Why This Matters Beyond the US-China Tech War
For African tech leaders and Nigerian businesses watching global AI dynamics, this plan has practical implications. It signals that within five years, there will be at least two meaningfully distinct AI computing ecosystems globally — one anchored by US hyperscalers and chip suppliers, another built around Chinese sovereign infrastructure. Which models, APIs, and cloud services your business integrates with today may increasingly reflect a geopolitical choice, not just a technical one. Nigerian enterprises building on AI now should begin to understand which vendors sit in which ecosystem — because that distinction will shape pricing, access, and policy exposure as both systems mature.
The Bigger Pattern: Sovereignty as Strategy
China's plan is the most dramatic example of a global trend — countries and regions treating AI infrastructure the way they once treated nuclear capability or satellite systems: as sovereign technology that cannot be outsourced. The EU's AI Gigafactories, India's IndiaAI Mission, and Nigeria's own emerging digital sovereignty conversations are smaller iterations of the same logic. The era of building AI applications entirely on top of foreign cloud infrastructure, with no domestic computing layer, is ending faster than most policymakers have noticed.
What do you think — should African nations be drafting their own AI infrastructure sovereign plans, or is cloud-leasing from US and Chinese hyperscalers the smarter near-term bet?
Originally featured on Tom's Hardware
