China's $295 Billion AI Data Centre Grid Has a Supply Problem That Money Cannot Fix

China is drafting a plan to spend roughly 2 trillion yuan, equivalent to $295 billion, over five years on a nationwide web of artificial intelligence (AI) data centres. The blueprint, reported by Bloomberg, requires that at least 80% of the underlying technology, AI chips included, come from domestic suppliers such as Huawei.

The National Development and Reform Commission is responsible for the network's design. State carriers China Mobile and China Telecom will operate most of the facilities and connect them into a single computing grid by 2028. Financing leans on sovereign debt and ultra-long special government bonds. When power grid upgrades are factored in, the total capital requirement could exceed 5 trillion yuan.

The construction ambition is not in question. The supply chain is.

The 80% domestic sourcing requirement effectively locks out chips from Nvidia and AMD, meaning the entire programme is capped by what China's Semiconductor Manufacturing International Corporation (SMIC) can physically produce. SMIC's most advanced stable node remains its N+2 process, roughly equivalent to 7 nanometres, and it is currently running above 93% utilisation. That leaves almost no headroom as every government-certified Chinese chipmaker competes for the same wafer slots.

High-bandwidth memory presents a separate bottleneck. Limited domestic supply of this specialist memory constrains how many Ascend-class accelerators Huawei can actually assemble. Huawei shipped around 812,000 chips last year and projects around $12 billion in processor revenue for 2026, a pace its own supply chain has struggled to sustain. Analysts estimate 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.

Beijing has meanwhile been tightening controls on foreign silicon. Last August it introduced a requirement that data centres source at least 50% of chips locally. 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.

China's own industry has expressed doubts about the pace. SMIC's co-chief executive Zhao Haijun cautioned that the rush to add capacity risks leaving data centres idle, comparing it to building highways ahead of traffic. Chinese chip executives have separately acknowledged the country trails the global leading edge in AI data centre silicon by between five and ten years. When AI company DeepSeek was directed toward Huawei hardware for model training, it reportedly reverted to Nvidia hardware after repeated difficulties, suggesting domestic parts still struggle with the heaviest training workloads.

The programme represents one of the largest planned infrastructure build-outs in recent memory. Whether the physical data centres get built on schedule is almost a separate question from whether they can be filled with capable enough hardware to justify the investment.