Dive Brief:

• Enterprise cloud storage spending will more than double to $128 billion by 2028, up from $57 billion last year, according to Omdia research. The firm analyzed IaaS and PaaS data center revenues for its annual Storage Data Services report.
• AWS dominated the storage segment, collecting 30% of total revenue, followed by Microsoft’s 22% slice and Google Cloud’s 14%. Amazon’s cloud division commanded a 38% share of capacity consumed, thanks to its massive lead in object storage, the most ubiquitous data repository in cloud, Dennis Hahn, principal analyst at Omdia, told CFO Dive sister publication CIO Dive.
• Hyperscaler storage revenue shares were roughly in line with the most recent quarterly split of total cloud services revenue, an IT spending category that will approach $700 billion this year, according to Gartner’s latest forecast. Omdia expects cloud storage spending to reach nearly $70 billion in 2024.

Dive Insight:

Enterprises entered an optimization cycle last year, trimming back cloud spend as the economy cooled. Storage services took a particularly hard hit, due to overprovisioning in prior years and a shift to pricey graphics processing units to fuel the emerging generative AI boom.

“There was a GPU gold rush that hurt storage spend quite a bit,” Hahn said. “If your out-of-pocket is going towards buying these really expensive GPU services, there’s less left for storage, networking and a lot of other things.”

The industry also absorbed the downstream impacts of a cloud buying spree, which drove a 30% year-over-year storage spend increase in 2022, compared to just 10% growth last year, according to Omdia’s research.

This year Omdia expects the market to grow 18% year over year as enterprises double down on the data side of AI-related technologies.

“Overall, analytics and generative AI are going to be a real boon to cloud — analytics already has been,” Hahn said. “As much as 70% of an AI project is combing through the data, getting it correct, cataloging it and trying to sort through it.”

AI adoption is driving shifts in cloud storage consumption patterns, as enterprises bulk up on data to feed large language model tools and applications. “Inferencing is going to have a lot larger impact on storage than LLM training,” Hahn said.

Not all storage is the same. Object storage dominates cloud, with 70% of total capacity, according to Omdia’s analysis, but only accounts for roughly one-third of storage spending. It is the cheapest storage type, most commonly used for data management and warehousing.

Block and file storage each comprise about one-third of cloud storage capacity and are pricier and more highly specialized. Block storage is optimized for databases, business intelligence systems and containerized workloads, while file supports office collaboration tools and server farm environments.

Hahn said he expects generative AI to accelerate file storage consumption. It is the preferred storage framework for retrieval-augmented generation, a common method for tuning LLMs to specific domains. It is also the most expensive cloud storage option.

AWS is ramping up file services, according to the report, for legacy application support and as a high-performance storage option for unstructured data interactions.