Beijing is building a national AI computing grid at a cost of $295 billion, and it is designed to keep American technology out.
The plan, led by the National Development and Reform Commission, calls for roughly 2 trillion yuan ($295 billion) over five years to build a nationwide network of interconnected AI data centers. Bloomberg reported the draft blueprint in early June, citing people familiar with the discussions. State-owned China Mobile and China Telecom would operate the bulk of the facilities and maintain the connections between them.
The most important detail is not the price tag. It is the rule that comes with the money.
At least 80% of the hardware and software used across the network, including AI chips, must come from Chinese suppliers. That means Huawei Technologies, the company already sanctioned by Washington for security violations, becomes the primary beneficiary. Nvidia and AMD, the two American firms that dominate the global market for high-end AI processors, are effectively locked out.
The 80% domestic sourcing threshold is not a suggestion. It is a procurement mandate written into the plan itself. For a company like Nvidia, which reported $19.7 billion in China revenue in the fiscal year ending January 2026, the effect is already visible. In the quarter ending April 2026, Nvidia shipped zero Data Center Hopper products to China, compared to $4.6 billion in the same period a year earlier. The Chinese AI chip market has been legislated out of reach.
Funding for the buildout will come from sovereign debt instruments, including ultra-long-term special government bonds, state-backed industry funds, commercial loans, and private investment. The data center push is one strand of China’s broader “Six Networks” program announced earlier this year, covering water, power, and computing infrastructure. When energy grid upgrades are included, the total cost could reach 5 trillion yuan ($735 billion).
The five-year plan running through 2030 mentions “AI” more than 50 times across 141 pages. It includes an “AI+ action plan” designed to embed artificial intelligence into industrial policy at every level. Vsevolod Smirnov, an analyst at Just2Trade, put it plainly: Chinese leaders need government projects to play catch-up with Western firms like Nvidia and Alphabet, each worth over $4 trillion in market cap.
There are signs the gap is narrowing.
Zhipu AI, the Beijing-based company behind the GLM series of large language models, recently surpassed HK$1 trillion ($128 billion) in market capitalization on the Hong Kong Stock Exchange. The company’s open-source GLM-5.2 model scored 74.4 on the FrontierSWE long-horizon coding benchmark, compared to Anthropic’s Opus 4.8 at 75.1 and OpenAI’s GPT-5.5 at 72.6. Independent firm Artificial Analysis ranked GLM-5.2 as the highest-rated open-source model ever on its Intelligence Index. Zhipu’s first post-IPO quarterly revenue hit 724.33 million yuan ($104.8 million), up 131.9% year over year.
Chinese data center operators are also expanding fast. Citi analysts identified GDS Holdings, which booked 200 megawatts of new capacity, and Vnet Group, with over 500 megawatts in new bookings, as well positioned to benefit from the national buildout.
The question is not whether Chinese AI will catch up. The analysts who follow these numbers say it is a matter of when, not if. The NDRC’s data center plan is the most aggressive state-directed infrastructure program China has attempted in this sector, and it comes with the full weight of Beijing’s policy apparatus behind it.
The $295 billion figure only covers public spending. Private-sector outlays from Alibaba, Tencent, and other Chinese tech giants are separate and could push the real number much higher. For comparison, American firms Meta and Microsoft have committed more than $700 billion for AI spending in 2026 alone. But the American model is corporate. The Chinese model is state-driven, centrally planned, and built to exclude foreign competition at every level.
That is not a bug in the plan. It is the plan.