China unveils $295 billion plan to build national AI data centers on homegrown chips by 2028

China Commits $295 Billion to Build AI Self-Reliance Through National Data Center Network

China is preparing to spend approximately 2 trillion yuan, or $295 billion, over the next five years on constructing a nationwide web of China AI data centers, according to sources familiar with the discussions 1

1

2

2

. The ambitious blueprint, led by the National Development and Reform Commission, aims to create an interconnected computing grid by 2028, with state-owned telecommunication companies China Mobile and China Telecom operating most of the computing facilities 2

2

3

3

. This represents Beijing's most aggressive AI infrastructure project yet, designed to challenge U.S. dominance in the intensifying global AI race.

The $295 billion plan forms a key component of China's broader "Six Networks" program, which encompasses computing, water, communication, urban underground pipe and logistics networks, and power grids 4

4

. When power grid upgrades are factored in, the total capital requirement could exceed 5 trillion yuan, sources told Bloomberg 1

1

. The funding will come largely from sovereign debt, including ultra-long special government bonds, plus state funds designated for strategic industries, supplemented by bank loans and private capital 3

3

4

4

.

Domestic Sourcing for AI Infrastructure Aims to Squeeze Out Nvidia and AMD

The most striking aspect of the blueprint is its requirement for domestic sourcing: at least 80% of the underlying technology, including AI chips, must come from local suppliers such as Huawei Technologies 1

1

2

2

. This domestic sourcing requirement effectively locks out Nvidia and AMD accelerators from the national grid of AI data centers, marking a deliberate effort to reduce reliance on foreign technology 3

3

5

5

. The move reflects the same sovereignty logic now sweeping the West, where Europe and Britain are scrambling to cut reliance on U.S. clouds, only inverted 3

3

.

Huawei's Ascend chipsets have demonstrated significant advancements, positioning them as a viable alternative to Nvidia's semiconductor products 5

5

. The company has outlined plans to launch a new AI chip annually over the next three years, with each iteration expected to offer double the computing power of its predecessor 5

5

. This timing reflects growing Chinese confidence in homegrown technology, particularly after nine domestic AI chips from Huawei, Alibaba, Shanghai Biren, and Moore Threads cleared a domestic security review in May, opening them to sensitive sectors 3

3

.

Domestic Chip Production Capacity Threatens to Derail Ambitious Timeline

While funding the build-out appears manageable, filling the facilities with domestic accelerators presents a fundamentally different challenge. The 80% domestic sourcing requirement means China will be capped by whatever amount of chips SMIC can physically produce 1

1

. The foundry's most advanced stable node remains its N+2 process, roughly equivalent to 7nm, which is currently running above 93% utilization, leaving little headroom as every government-certified Chinese chipmaker competes for the same wafer slots 1

1

.

Another major chokepoint is high-bandwidth memory. Highly limited domestic HBM supply constrains how many Ascend-class accelerators Huawei can assemble 1

1

. Huawei alone shipped around 812,000 chips last year and projects some $12 billion in processor revenue for 2026, a pace that its own AI supply chain has struggled to sustain 1

1

. It's estimated that China's domestic suppliers will cover only around 76% of all Chinese AI chip demand by 2030, even as that market grows toward $67 billion 1

1

.

China's AI Capabilities Face Reality Check Despite Industry Optimism

China's own industry has questioned whether domestic hardware can keep pace with the ambitious rollout. SMIC co-CEO Zhao Haijun has cautioned that the rush to add capacity risks leaving data centers idle, comparing it to building highways ahead of the traffic 1

1

. Chinese chip executives have separately conceded the country trails the leading edge in AI data center silicon by five to 10 years 1

1

. When DeepSeek was steered toward Huawei hardware for model training, it eventually reverted to Nvidia hardware, lending credence to the idea that domestic parts still struggle with the heaviest training workloads, even where they suffice for inference 1

1

.

Beijing has massively tightened its restrictions on foreign silicon through a series of new controls. Last August, Beijing introduced a requirement that data centers source at least 50% of chips locally, and by November, state-funded projects were barred from foreign accelerators entirely, with builds less than 30% complete reportedly told to strip out Nvidia, AMD, and Intel parts 1

1

. These semiconductor restrictions have shaped China's strategic pivot toward AI self-reliance.

Context and Scale Reveal Different Strategic Calculus

The $295 billion figure, while substantial, is spread over five years and appears modest compared to Western spending. U.S. firms alone, led by Meta and Microsoft, are setting aside roughly $725 billion for AI this year 3

3

. Meta has a planned capital expenditure spend of as much as $145 billion for this year alone, while Alphabet has set aside up to $190 billion 4

4

. However, the Chinese government spend doesn't account for private AI investment in China, which amounted to around $12.4 billion in 2025 4

4

. Chinese AI darling DeepSeek is reportedly nearing a $7.4 billion raise backed by the likes of Tencent, while Alibaba led a $293 million funding round into ShengShu Technology 4

4

.

China's core AI industry, which boasts more than 6,200 companies, was valued at nearly $174 billion in 2025, according to a government statement from March 4

4

. International Data Corporation placed the Chinese AI market at some $63 billion at the end of 2025, with estimates expecting it to cross $200 billion by 2029 4

4

. The point is less the raw sum than the coordination: a state marshalling debt, land, power, and chips behind a single national grid 3

3

. "Elevating it to a national strategy ensures policy alignment and capital mobilisation," said Charlie Dai, a principal analyst at Forrester 3

3

. The plan remains in early discussions and details could change, but the direction signals that the world's two largest economies are each trying to wall off their own AI supply chains, and the era of a single global stack is ending 3

3

.