China's AI Computing Capacity to Triple with 8.5 Trillion Yuan Investment, Domestic Chips to Exceed 70% Share

According to a report, Goldman Sachs' latest AI investment landscape research indicates that between 2026 and 2030, China's total AI-related capital expenditure will reach 8.5 trillion yuan, averaging 17% of the total capital expenditure of comparable US enterprises.

This massive investment will support a significant expansion of the nation's IT power capacity.

The report projects that China's total IT capacity will rise from 26GW in 2025 to 81GW in 2030, more than tripling in size, with new AI power demand growing from 5.6GW in 2026 to 13GW in 2030.

Capital expenditure is composed of three main pillars.

The first pillar is hyperscale cloud providers and major internet companies, with capital expenditure expected to reach 995 billion yuan in 2026, a year-on-year increase of 121%, far exceeding the 41% growth of US peers, with Alibaba, Tencent, and Baidu as the primary participants.

The second pillar is new cloud providers, with total investment over the next five years expected to reach 1.5 trillion yuan.

Investment in 2026 is projected at 234 billion yuan, rising to 363 billion yuan by 2030, with more than 95% allocated to server procurement.

The third pillar is telecom operators, with total computing power capital expenditure from 2026 to 2030 expected to reach 653 billion yuan, representing a compound annual growth rate of approximately 10%.

In 2026, the three operators combined will account for approximately 81 billion yuan, with China Mobile contributing 38 billion yuan, China Telecom 26 billion yuan, and China Unicom 18 billion yuan.

Domestic chips are the core beneficiary of this expansion round.

The report estimates that domestic AI chip shipments will grow from 1.1 million units in 2025 to 2.4 million units in 2026, 4.8 million units in 2027, and reach 11 million units by 2030, with market size expanding from 94 billion yuan to 646 billion yuan.

The deployment share of domestic chips in servers will rise substantially.

The report predicts that domestic chip deployment in servers will reach 70% to 85%, meaning domestic GPUs and ASICs will occupy a dominant position in this round of computing capacity expansion.