Firms chase AI ROI while cloud storage costs and security risks mount
Many firms are pouring money into AI infrastructure even though most of their projects have yet to turn a profit, according to the fourth annual Wasabi Global Cloud Storage Index.
The study, developed with Vanson Bourne and based on responses worldwide, finds that just 32 percent of respondents say their AI projects are delivering a positive return today.
Expectations shift quickly, however: 51 percent expect AI projects to generate positive ROI within the next 12 months, a jump of 19 points according to the research.
Despite the limited return so far, AI infrastructure spend is “a corporate priority.”
When asked about infrastructure budgets for AI, only 3 percent of respondents say they will lower spending, 60 percent plan to increase it, and 37 percent intend to keep it unchanged, as per the Wasabi Cloud Storage Index.
Budget allocation patterns also differ from the broader public cloud market.
According to the Index, approximately two-thirds of AI budgets (66 percent) go to data, storage, and processing power, while 33 percent is allocated to AI software and SaaS solutions.
Andrew Smith, director of strategy and market intelligence at Wasabi Technologies and a former IDC analyst, said that “the vast majority” of public cloud services revenue still comes from software/SaaS rather than infrastructure services.
But he said emerging AI workloads are now “changing this dynamic… it’s the complete opposite of what we might expect from a traditional market standpoint.”
The research highlights data storage and data quality as the main operational hurdles.
Respondents identify data storage (including cost, access, management) as their top challenge in implementing AI projects and solutions, followed by data quality (including cleansing, preparation).
Compute challenges (including cost, procurement) rank third, indicating that data and storage concerns outweigh processing issues in many AI deployments.
Hybrid storage strategies have become common as organizations try to manage AI-related data across multiple environments.
According to the survey, 64 percent of respondents are deploying hybrid storage solutions that mix on-premises and public cloud to support AI workflows.
Respondents say they prefer public cloud storage for two specific workflow stages: data retrieval/ingest and aggregation, and model retention and archiving, effectively using public cloud storage to “bookend” the AI data pipeline.
Cost pressures from cloud storage fees remain persistent.
For the fourth year in a row, the Cloud Storage Index finds that about 50 percent of user spending on public cloud storage goes to fees rather than storage capacity.
This fee-heavy structure feeds into budget overruns: 49 percent of respondents say they exceeded budgeted spending for cloud storage in 2025, driven by a combination of increased storage usage, growth, migration, and fees.
In total, 91 percent of respondents cite at least one fee-related reason why their spending on public cloud storage exceeded budget expectations, according to the report.
Security adds another layer of risk.
The findings show that 44 percent of respondents have experienced a cyberattack that resulted in loss of access to public cloud data.
At the same time, 41 percent say their public cloud vendor does not provide the necessary tools and features needed to mitigate against cyberattack, a gap the 2026 Cloud Storage Index flags as an important market issue.
