๐Bloomberg TechnologyโขStalecollected in 21m
China Announces $295 Billion Plan for AI Data Centers
๐กA massive $295B infrastructure push that will reshape the global demand for AI compute and data center hardware.
โก 30-Second TL;DR
What Changed
Investment of 2 trillion yuan dedicated to national data center expansion.
Why It Matters
This massive infrastructure investment will significantly increase the demand for high-performance computing hardware and energy-efficient data center technologies.
What To Do Next
Evaluate your supply chain dependencies for hardware and cooling solutions as global demand for data center infrastructure surges.
Who should care:Developers & AI Engineers
Key Points
- โขInvestment of 2 trillion yuan dedicated to national data center expansion.
- โขFive-year timeline for infrastructure deployment to support AI workloads.
- โขStrategic move to bolster domestic computational capacity for AI development.
๐ง Deep Insight
Web-grounded analysis with 18 cited sources.
๐ Enhanced Key Takeaways
- โขThe 2 trillion yuan ($295 billion) investment is part of a broader "Six Networks" national infrastructure strategy, and the total projected investment could reach at least 5 trillion yuan (approximately $738 billion) when power grid integration is included.
- โขThe plan mandates that at least 80% of the core technologies, including AI chips, must be sourced from domestic suppliers like Huawei, aiming to reduce reliance on foreign companies such as Nvidia and AMD.
- โขThe initiative aims to connect China's currently fragmented public and private data facilities into a unified national computing network by 2028, treating computing power as a public utility.
- โขThis investment builds upon the existing "Eastern Data, Western Computing" (EDWC) project, launched in 2022, which strategically relocates data processing to energy-rich western regions to optimize energy use and promote regional development.
- โขState-owned telecommunication companies, including China Mobile and China Telecom, are designated to operate the majority of these new data centers and ensure their interconnection.
๐ ๏ธ Technical Deep Dive
- Domestic Chip Reliance: The plan emphasizes using domestic AI chips, with nine Chinese-made AI chips from companies like Huawei, Alibaba, Birun Technology, and Moore Threads having passed national security evaluations in May 2026.
- Hyperscale Infrastructure: AI data centers are optimized for deep learning, large language models, and high-performance computing, integrating specialized hardware like GPUs, AI accelerators, and high-bandwidth networking.
- Advanced Cooling Systems: Facilities are adopting liquid cooling technologies and AI-powered energy management systems. For instance, a mega data center in Inner Mongolia utilizes a liquid cooling system circulating 20,000 gallons of coolant per minute.
- Network Architecture: The infrastructure features high-speed internal bandwidth, with one mega data center boasting 400 terabits per second of internal bandwidth and over 6,000 meters of fiber optic cable to ensure microsecond communication between GPUs.
- Energy Efficiency & Renewables: The "Eastern Data, Western Computing" project targets over 80% green electricity usage in new data centers and aims for Power Usage Effectiveness (PUE) below 1.25 for eastern hubs and 1.2 for western hubs.
- Underwater Data Centers: China has deployed innovative solutions like the world's first wind-powered underwater data center off Shanghai, which has a 24-megawatt capacity and reduces power consumption by over one-fifth due to natural seawater cooling.
- Low-Latency Networks: The Lingban Computing Network platform by China Mobile Hubei provides multi-level latency coverage, achieving 1-millisecond latency for intra-city access and 10 milliseconds to national core computing hubs.
๐ฎ Future ImplicationsAI analysis grounded in cited sources
China will significantly increase its technological sovereignty in AI hardware.
The mandate for 80% domestic technology, including AI chips, will accelerate the development and adoption of indigenous AI hardware, reducing reliance on foreign suppliers.
The cost of AI computing power in China will decrease for businesses and researchers.
By establishing computing power as a national public utility and optimizing resource allocation through a unified network, China aims to lower the cost and usage threshold of computing power.
China's western regions will see accelerated economic development and job creation.
The 'Eastern Data, Western Computing' strategy directs significant infrastructure investment and data processing workloads to the western provinces, fostering new industries and employment opportunities.
โณ Timeline
2006
AI included in National Medium and Long Term Plan for the Development of Science and Technology and 11th Five-Year Plan.
2016
13th Five-Year Plan sets a goal for China to become a global AI leader by 2030.
2017
New Generation Artificial Intelligence Development Plan released, aiming for a $150 billion national AI industry.
2022-02
The 'Eastern Data, Western Computing' (EDWC) national megaproject officially launched.
2024-07
Guidelines for the Construction of a Comprehensive Standardization System for the National Artificial Intelligence Industry introduced.
2026-05
Nine Chinese-made AI chips pass national security evaluations, clearing way for widespread domestic deployment.
๐ Sources (18)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
๐ฐ
Weekly AI Recap
Read this week's curated digest of top AI events โ
๐Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: Bloomberg Technology โ