WD says AI-generated video leaves up to 130 times more data in the back end
🎧 Listen to this story
AI inference runs continuously and leaves behind logs, cached context, immediate renders, backups, and safety records. According to internal WD measurements cited by MIT Technology Review, an eight-second clip returned as a 3.97MB 720p file generated more than 500MB of associated back-end data—130 times more; longer, higher-resolution videos can reach hundreds of gigabytes per clip.
McKinsey projects that inference demand will grow at a 35% compound annual growth rate through 2030, versus 22% for training, and account for more than 40% of total data-center demand by the end of the decade. In a WD-sponsored global IDC white paper, nearly 95% of 763 respondents said their data volume had increased because of generative AI over the previous 12 months, while 96% expected to need faster archive retrieval for inference or retrieval-augmented generation (RAG).
Why it matters · editorial interpretation
The gap between the delivered file size and the data retained in the back end could make storage, archiving, and retrieval central factors in the cost and planning of video inference. McKinsey’s projections and IDC’s data suggest that this pressure could expand as demand for inference and RAG grows.
Sources
Hardware