Intelligent World 2030

Data Storage 2030

By 2030, we will be producing yottabytes of data, and advancements in data storage technology will drive human civilization to new heights. Building new data infrastructure can accelerate the transformation towards an intelligent society, which will enable us to understand the world more deeply and unlock the potential of a brighter future.

Download Report

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

1 YB of data is expected to be generated globally each year, and more than 80% of it will be unstructured

100% of mass hot data will be stored on SSDs

More than 60% of enterprises will need to be able to access active archived data at least once a day

More than 80% of cloud and Internet enterprises will use the diskless reference architecture

More than 75% of endpoint data, 80% of edge data, and 90% of data in core data centers will be processed by AI in real time

Over 80% of enterprises are expected to deploy multi-layer ransomware protection systems which cover the storage systems

Key Challenges

Key Technical Features

Advanced Media Application

Data tiering: shifting from three tiers—hot, warm, and cold data—to two tiers—a hot tier and a warm/cold-combined tier.
Advanced media technology: storage media for hot data—higher capacity, lower power consumption, and higher concurrency; storage media for warm data—HDDs are the mainstream; storage media for cold data—high reliability, long lifespan, and high environmental adaptability.
Media application innovation: Wafer-scale and chiplet technologies are becoming the mainstream. New data coding technologies further improve capacity density.

Show Less

Data-centric Architecture

Decoupling storage and compute resources at the macro level and enhancing compute with storage: Storage resource pooling, memory-based storage that supports global memory semantic access, and NPUDirect Storage enable decoupled and flexible scheduling of compute and storage resources, maximizing resource utilization.
Coupling storage and compute resources at the micro level and eliminating repeated computing through queries: Near-data processing reduces unnecessary data movement and repeated computing by 90%, improving data processing efficiency.
Cluster storage: Scalable cluster storage can be scaled out to thousands of nodes and scaled up to millions of xPUs, with all xPUs working simultaneously.

Show Less

Intrinsic Data Resilience

Proactive data protection: building a proactive data protection resilience system from multiple technical directions, such as data security situational awareness, data timeline travel, native anti-tampering, and multi-dimensional linkage response.
Zero data copy: breaking down data boundaries and enabling data sharing through utilization on zero data copy access technology.
Zero-trust storage: resolving resilience issues through technical breakthroughs in full-path data encryption, mandatory data access control, and privacy computing.

Show Less

Intelligent Data Fabric

Automatic data orchestration: developing data profiles and data brains to achieve optimal data placement without compromising on service performance.
Cross-region data collaboration: providing unified compute-storage services for lower costs and better infrastructure capabilities.
Storage network: constructing an efficient and fast storage network to implement data access that is insensible to applications and regions.

Show Less

Data Intelligence

Service interface for content consumption: evolving from simple data access to content consumption to enable long-term memory storage for AI.
Data semantics extraction: compressing the original data and improving system efficiency.
Multi-modal data analysis: standardizing and processing data from different sources so that the data can be exchanged and shared between different applications.
Data adaptive modeling: automatically identifying and learning patterns and structures from data as it is ingested and generating corresponding prediction models.

Show Less

Sustainable Storage

Storage system–level energy saving: using models to adjust hardware and software working status parameters to achieve optimal energy consumption of the entire system.
Data transmission energy efficiency improvement: making breakthroughs in nanosecond-level optical switching technology and high-speed switching algorithm to achieve an all-optical data center network with low power consumption.
Chip-level energy saving technologies: utilizing technologies such as heterogeneous and diversified computing power integration and on-chip dynamic intelligent energy efficiency management to realize both high computing power and low power consumption.
Green and intensive storage standards: setting unified green and intensive standards that cover key data indicators and creating an energy consumption benchmark from which a comprehensive evaluation system will be established for a green and low-carbon storage industry.

Show Less