The effective AI computing power provided by a single cluster is expected to reach 100 EFLOPS
Data Center 2030
Indicator prediction
Continuous innovation in computing power supply and breakthroughs in challenging resource constraints will become the main theme of data center development in the future
All-flash storage is expected to account for 80% at data centers
The automation level of industry-leading data centers is expected to reach L4
The green power usage of large data centers is expected to reach 100%
Function-level computing, storage, and network resource allocation is expected to be possible in leading cloud data centers
The penetration rate of hyper-converged Ethernet in data centers is expected to reach 80%
Key features
Improve the overall efficiency of data centers through systematic innovation, maximize the scissors difference between computing power supply and resource constraints, and accelerate towards an intelligent world
Diverse and ubiquitous
- Big clusters: construction from a single server to a cabinet; O&M from a single server to a cluster or even the entire data center
- Lightweight edges: from the center to the edge, meeting low-latency service processing and data security requirements of various industries
- New patterns: construction of underwater data centers and space data centers to meet various application scenarios
Secure and intelligent
- High security: new security ecosystem in the era of trusted computing, confidential computing, and AI foundation models
- High reliability: cross-domain high reliability in a larger scope and service-level high reliability that focuses on service continuity
- High intelligence: AI and data enabling the full lifecycle of data center planning, construction, and operation
Zero carbon and energy saving
- Green power supply: wind power, solar power, and hydropower, and dynamic microgrid
- New energy storage: from direct power supply to power grid + energy storage system
- Liquid cooling: full liquid cooling and air-liquid hybrid cooling for low carbon throughout the lifecycle, achieving an optimal PUE
Flexible resources
- Full pooling: application-centric; pooling of computing, transport, and storage resources
- Flexible computing: providing flexible computing services based on utilization and QoS; shortening the computing lease term from days to hours or minutes
- Ubiquitous collaboration: from a single region to cross-region synergy
Peer-to-peer interconnection
- Hyper-convergence: multiple computing buses and network technologies converged and unified to implement heterogeneous peer-to-peer passthrough
- High performance: innovative design of next-generation high-performance computing data centers based on the high-speed peer-to-peer interconnection architecture
- Optical-intrinsic: all-optical interconnection within data centers
More Industry Reports
-
Communications Network 2030
Learn More -
Computing 2030
Learn More -
Data Storage 2030
Learn More -
ICT Services & Software 2030
Learn More -
Cloud Computing 2030
Learn More -
Digital Power 2030
Learn More -
Intelligent Automotive Solution 2030
Learn More -
Intelligent Campus 2030
Learn More -
Electric Power Digitalization 2030
Learn More -
Green Development 2030
Learn More