A global study by IDC reveals that AI infrastructure economics are shifting as enterprise data accumulates and persists over time. The white paper, titled “Built for Scale: The Enduring Role of HDDs in the AI Era” and sponsored by Western Digital, examines how artificial intelligence creates structural expansions in storage needs across organizations.
Surge in Data Volumes Across Enterprises
The quantitative survey encompassed 763 IT and business decision-makers across seven countries, including Saudi Arabia, the United States, China, Japan, South Korea, India, and Germany. Findings show that 94.7% of surveyed organizations store more data because of AI and generative AI adoption over the past 12 months. In addition, 61% experienced data growth of 25% or more in the last year, while 74% expect volumes to increase by 25% or more over the next three years.
Data lakes reflect this upward trajectory directly, with 85.4% of respondents reporting volume growth. Specifically, 59.4% identified AI-generated inputs, such as synthetic data, inference outputs, and model logs, as the primary driver of this expansion.
AI infrastructure economics in Practice
Unlike compute cycles that terminate once processing completes, AI data remains stored for future model cycles. As a result, the dynamics of AI infrastructure economics require IT leaders to balance compute investments with long-term data management budgets. Nearly 95% of organizations stated that the value of their data increased following generative AI implementation, and 74.3% reported retaining data for longer periods.
“For the last few years, the AI infrastructure conversation has centered on compute. But AI runs on data. Organizations are generating more data, keeping it longer, and finding new ways to create value from the information they already have.”
Irving Tan, CEO, WD
Reactivating Cold Storage for Workloads
The study highlights a change in traditional data classification, as historical archives re-enter operational environments. Around 75.9% of organizations retrieve cold-tier data to support new AI workloads, including retrieval-augmented generation. Furthermore, 96% anticipate needing faster archive retrieval speeds for future inference applications.
Within surveyed enterprise architectures, 74.6% of data sits in warm, cool, and cold tiers, while cold data constitutes over 60% of data lake capacity. Consequently, organizations are adjusting their cybersecurity and storage retention strategies to maintain rapid access to older data.
Strategic Balancing of Enterprise Storage
Managing long-term data stores effectively now represents a core requirement for enterprise economy plans. The survey notes that 98.2% of organizations consider the total cost of ownership per terabyte an important or very important factor in storage decisions. Managing storage capacity alongside compute resources will shape how efficiently enterprises deploy advanced models over the coming decade.
