A global study sponsored by WD shows that as data continues to accumulate and grow exponentially, the economic model of AI infrastructure is being reshaped. According to the latest IDC survey, AI is not only driving a rapid increase in data volume but also extending data retention times and reactivating demand for historical data.
AI-Driven Data Growth
This study, conducted in Hong Kong, found that AI is driving structural growth in data storage needs. Among the organizations surveyed, 94.7% reported that their stored data volume has continued to increase over the past 12 months due to the widespread adoption of AI and generative AI. 61% of the surveyed organizations experienced a data growth of 25% or more and anticipate that data volume will continue to grow over the next three years.
Additionally, 85.4% of the surveyed organizations stated that their data lake capacity has been expanding over the past year, and 59.4% of the organizations believe that AI-generated data is the main factor driving the growth of data lake capacity.
Extended Data Retention Times
The survey also indicated that nearly 95% of the surveyed organizations reported an increase in the value of their data after adopting AI. 74.3% of the surveyed organizations have extended their data retention times, and 75.9% of the organizations are bringing more archived cold data back online to support AI workloads.
According to the study, 96% of the surveyed organizations anticipate the need to retrieve archived data more quickly to support AI inference and retrieval-augmented generation applications.
AI Infrastructure Design Requirements
WD CEO Irving Tan stated that AI operates based on data, and the demand for large-scale storage, management, and access to data will continue to grow. This data foundation will play a key role in determining the future potential of AI.
The study points out that AI workloads generate data throughout their lifecycle, and this data continues to be retained even after the related workloads are completed, potentially serving as input data for future AI applications. As data volume grows, retention times extend, and historical data is reactivated, organizations need to balance performance, capacity, accessibility, and cost-effectiveness.
Conclusion
The ability to manage data on a large scale in a cost-effective manner is no longer a secondary consideration for infrastructure but is increasingly becoming a key factor in determining the success or failure of AI. Comprehensive data lifecycle design will be an important consideration for future AI infrastructure.