China Bans Foreign AI Chips from State-Funded Data Centers, Offers Power Subsidies for Domestic Alternatives

Reviewed byNidhi Govil

19 Sources

Share

China has issued sweeping restrictions on foreign AI chips in government-funded data centers while simultaneously offering substantial power subsidies to incentivize the use of domestic semiconductors. The move represents a significant escalation in the country's push for technological independence from U.S. chip manufacturers.

China Implements Comprehensive Ban on Foreign AI Chips

China has issued sweeping new restrictions on foreign AI accelerators in government-funded data centers, marking a significant escalation in the country's push for technological independence from U.S. semiconductor manufacturers. According to multiple reports, the directive requires all state-backed data center projects to use only domestically produced AI chips going forward

1

4

.

Source: Seeking Alpha

Source: Seeking Alpha

The new guidance applies retroactively to construction projects that are less than 30% complete, meaning accelerators from Nvidia, AMD, and Intel may need to be removed or replaced if already installed. Projects at more advanced stages will be assessed on a case-by-case basis, though the directive effectively eliminates prospects for foreign chip manufacturers in China's government-backed AI infrastructure

1

.

Power Subsidies Incentivize Domestic Chip Adoption

To offset the performance disadvantages of domestic semiconductors, Chinese provinces are offering substantial electricity subsidies to data centers using homegrown chips. Local governments in Gansu, Guizhou, and Inner Mongolia are providing discounts of up to 50% on industrial power pricing, bringing costs down to approximately 0.4 yuan (5.6 cents) per kilowatt-hour

2

3

.

Source: Wccftech

Source: Wccftech

These subsidies address a critical weakness in China's domestic chip ecosystem. Current-generation Chinese AI accelerators require 30 to 50% more electricity to generate the same computational output as Nvidia's H20 chips, according to industry experts. The energy efficiency gap becomes even more pronounced when compared to Nvidia's newer Blackwell architecture, which remains subject to U.S. export controls

2

.

Impact on Major Tech Companies and Market Dynamics

The directive affects China's largest technology companies, including ByteDance, Alibaba, and Tencent, which have complained to regulators about increased operational costs associated with domestic semiconductors. Some companies had received cash incentives substantial enough to cover a data center's operating costs for approximately one year, demonstrating the government's commitment to supporting the transition

2

.

Source: FT

Source: FT

For Nvidia, which once commanded over 90% of China's AI accelerator market, the restrictions represent a devastating blow to its Chinese operations. CEO Jensen Huang recently acknowledged that the company's market share in China has "effectively plummeted to zero"

1

. The ban encompasses even Nvidia's H20 chips, which were specifically designed to comply with U.S. export restrictions.

Strategic Implications and Domestic Chip Development

China has invested more than $100 billion in AI infrastructure projects over the past two years, giving Beijing's directive significant leverage over the market. The policy represents a formal shift from earlier approaches that merely discouraged foreign chip purchases to an outright prohibition backed by financial penalties

1

.

Domestic alternatives include chips from Huawei, Cambricon, and Enflame, with Huawei's Ascend line considered the most mature option. However, Chinese chipmakers continue to lag behind their U.S. counterparts in software tooling and performance density. Huawei has attempted to compensate for weaker single-chip performance by combining multiple processors into larger clusters, though this approach increases operational electricity costs

1

2

.

TheOutpost.ai

Your Daily Dose of Curated AI News

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Spend less time searching for the latest in AI and get straight to action.

© 2025 Triveous Technologies Private Limited
Instagram logo
LinkedIn logo