China Plans $295B AI Data Center Buildout to Rival the US

China is preparing to spend roughly 2 trillion yuan — about $295 billion — over the next five years building AI data centers across the country, according to a Bloomberg report published today. The plan is one of Beijing's clearest signals yet that it intends to challenge US dominance in artificial intelligence.

The scale is enormous, and so is the strategy behind it: the buildout is designed to run mostly on Chinese-made technology, effectively shutting American chipmakers Nvidia and AMD out of the world's second-largest AI market.

Why now

The timing is no accident. For two years, US export controls have steadily tightened the flow of advanced AI chips into China, leaving Chinese firms scrambling for compute. Rather than keep fighting for access to American hardware, Beijing appears to be choosing self-reliance.

Why this matters: compute — the raw processing power that trains and runs AI models — is the single biggest bottleneck in the AI race. A government willing to spend $295 billion to guarantee its own supply is making a long-term bet that it can build a parallel AI stack independent of the US.

The plan also fits a familiar Chinese playbook: pour state money into a strategic industry, coordinate it from the top, and lean on national champions to execute.

What's actually in the plan

According to the report, key agencies including the National Development and Reform Commission are drafting a blueprint to build a network of interconnected computing hubs nationwide.

State-owned giants China Mobile and China Telecom would operate the bulk of the data centers and stitch them together into a unified network. The goal is to connect China's currently scattered data infrastructure into one coordinated system by 2028.

The most consequential detail is the sourcing rule: the plan reportedly calls for at least 80% of the technology — including AI chips — to come from domestic suppliers like Huawei. That's the mechanism that squeezes out Nvidia and AMD.

Funding would come mainly through sovereign debt, including ultra-long-term special government bonds (typically more than 10 years in duration) plus state investment funds aimed at strategic industries. When power-grid upgrades are factored in, total projected spending could reach at least 5 trillion yuan.

Who this affects

The most obvious losers are Nvidia and AMD. Both have spent years navigating US restrictions to keep selling into China; a state directive favoring domestic chips for 80% of a national buildout would wall off a huge chunk of future demand.

The biggest winner is Huawei, which has been developing its own AI accelerators as an alternative to Nvidia's. A government-backed mandate to fill national data centers with domestic silicon is exactly the kind of guaranteed demand that could help Chinese chipmakers close the performance gap through sheer volume and iteration.

For the broader AI industry, the plan deepens a split that's been forming for a while: two increasingly separate AI ecosystems, one built on American hardware and one on Chinese hardware, with less and less overlap between them.

The skeptics' view

Not everyone is convinced money solves China's problem. Some analysts argue that capital alone can't overcome the country's chip constraints — that the gap between Huawei's accelerators and Nvidia's best parts is a matter of manufacturing and design maturity, not just spending. Throwing $295 billion at data centers doesn't automatically produce frontier-grade chips to fill them.

That tension — vast ambition meeting hard technical limits — is the central question hanging over the plan. A national network is only as strong as the chips inside it.

What's next

It's worth stressing this is a plan in the drafting stage, not a finished policy. The blueprint is still being shaped by government agencies, and details could shift before anything is finalized.

But the direction of travel is clear. Beijing is committing to a domestic-first AI infrastructure strategy and is willing to fund it at a scale that rivals the biggest US data-center programs. Watch for formal confirmation of the plan, the first wave of contracts to state operators, and any signs of how Huawei's chips perform at this scale.

The bottom line: the AI race is increasingly a race over infrastructure and chips, not just models. China just signaled it's prepared to spend nearly $300 billion to build its own — and to do it without American hardware. For anyone tracking where AI is headed, the contest between US and Chinese compute is now one of the defining storylines of the decade.