Data-driven Synergy: How Digitalization and Intelligentization Fuel Each Other
-
Finance
Enhancing multi-source, mass data management and strengthening data resilience compliance
-
Carriers
Activating mass data to facilitate the efficient training and the implementation of large AI models across industries
-
Public Services
Jointly streamlining cross-department data and protecting sensitive data to enhance public services
-
Manufacturing
Finding value in historical, dormant data and boosting E2E production efficiency
-
Electric Power
Promoting multi-dimensional and high-frequency data collection and secure data retention for more precise electricity supply and demand forecasting
-
Education and Research
High-performance, reliable, and resilient data supply underpins AI-driven intelligentization
-
Healthcare
Efficient and resilient data sharing protects patient privacy
Finance
Carriers
Public Services
Manufacturing
Electric Power
Education and Research
Healthcare
Digital and Intelligent Transformation Across Industries Requires High-Quality Data and Efficient Data Processing
Data Awakening
- Data awakening is a must for transitioning from the digital era to the intelligent era. During model training and inference, activating idle service data and waking up historical archived data help address the challenge of data shortage.
- Collecting and managing high-quality training data is a must for continuous AI evolution.
Data Generation and Synthesis
- Data generation and synthesis power the digital-intelligent era, effectively facilitating the rapid development of large AI models.
- Data generation: It explores five key dimensions to generate, collect, and retain high-quality data for AI—data generation/collection site, format, frequency, full-process service data, and future-proof data retention.
- Data synthesis: Synthetic data is a beneficial supplement to the raw data obtained in the real world. It can address key challenges like data scarcity and privacy protection.
Data Efficiency
- In the digital-intelligent era, data storage needs to be continuously optimized across six dimensions: ultra performance, scalability, data resilience, data fabric, new data paradigm, and sustainability. This is essential to fully enhance data efficiency and unleash data productivity.
Trends and Suggestions on Data Infrastructure in the Digital-Intelligent Era
Trend Analysis
1Multi-modal AI training is generating larger volumes of more complex data.
2AI computing clusters are expanding in scale but declining in computing power utilization.
3Hallucination is common in AI inference.
Suggestions
1Use the decoupled storage-compute architecture to enable independent deployment and on-demand evolution of computing and storage power.
2The data infrastructure should have scale-out capabilities, enabling performance to increase proportionally to capacity.
3The data infrastructure should support multi-protocol interworking.
