China Wants to Build a Massive $295 Billion AI Supergrid Using Locally Produced Chips

Image Courtesy: Getty Images

China is preparing an ambitious multi-year plan to build a nationwide artificial intelligence computing network, with authorities reportedly proposing investments of around 2 trillion yuan ($295 billion) to establish interconnected AI data centers powered primarily by domestically produced technology.

The initiative, being developed by the National Development and Reform Commission, would create a national computing grid linking facilities operated largely by state-owned telecom giants China Mobile and China Telecom. The network is expected to be completed by 2028, with at least 80% of the underlying hardware, including AI processors, sourced from Chinese suppliers such as Huawei.

The project represents one of China’s largest AI infrastructure pushes to date and forms part of a broader effort to reduce dependence on foreign technology amid ongoing export restrictions and geopolitical tensions surrounding advanced semiconductors.

While financing the construction appears achievable through sovereign debt and long-term government bonds, supplying enough domestic AI chips could prove far more challenging. The proposed 80% local sourcing requirement would largely exclude accelerators from Nvidia, AMD, and other foreign manufacturers, placing enormous pressure on China’s domestic semiconductor ecosystem.

Much of that burden falls on SMIC, China’s leading contract chipmaker. Industry analysts note that the company’s most advanced production technology remains several generations behind the world’s cutting-edge semiconductor processes, while factory utilization rates are already running at extremely high levels.

Memory technology presents another bottleneck. High-bandwidth memory, a critical component for modern AI accelerators, remains in limited supply domestically. That constraint could restrict how many advanced AI processors companies such as Huawei are able to manufacture, even as demand for computing power continues to surge.

The proposal follows a series of increasingly strict policies designed to prioritize Chinese hardware in strategic technology projects. Over the past year, Beijing has expanded local sourcing requirements for data centers and reportedly moved to phase out foreign accelerators from state-backed AI initiatives.

Despite rapid progress, industry leaders within China have publicly acknowledged that the country still trails global leaders in AI data center hardware. Some executives estimate domestic AI chips remain five to ten years behind the most advanced international offerings for demanding training workloads, although they may be increasingly competitive for inference and other specialized tasks.

If completed as envisioned, the network would become one of the world’s largest state-backed AI infrastructure projects, potentially reshaping the global semiconductor market while accelerating China’s push for technological self-sufficiency in artificial intelligence.