Companies worldwide are allocating growing sums to artificial intelligence, yet many struggle to measure the business returns precisely. According to IDC, global spending on AI infrastructure reached $318 billion in 2025, up from $153 billion in 2024. Experts point to fragmented cost structures and limited visibility as key reasons why ROI often remains unclear.
The scale of the increase
Deloitte’s 2025 research found that 85 percent of organizations increased their AI investments compared to the previous year, and 91 percent planned further spending increases. Corporate-level data from CloudZero shows a comparable trend: average monthly AI spend rose 36 percent year over year, from $62,964 to $85,521. The share of organizations spending more than $100,000 per month on AI tools grew from 20 percent to 45 percent.
Returns are uneven
Despite rising investment levels, the business impact of AI is not always straightforward to isolate. Deloitte notes that AI deployments are frequently rolled out alongside other digital, operational, or organizational changes, making it difficult to attribute outcomes directly to AI. The Boston Consulting Group’s survey from last year found that 60 percent of companies still report only limited business benefits from their AI investments or cannot demonstrate significant revenue increases or cost reductions.
Measurable gains in some areas
At the same time, a growing number of organizations report tangible returns. Google Cloud’s 2025 survey of 3,466 executives showed that 74 percent of respondents had seen ROI in at least one generative AI use case. The most commonly reported measurable benefits were improved individual productivity (39 percent), enhanced customer experience (37 percent), and support for sales and marketing activities (33 percent).
Fragmented costs complicate visibility
Measuring AI ROI is further complicated by the fact that related costs appear across multiple categories: model usage fees, token costs, GPU resources, cloud infrastructure bills, and integration and operational efforts. Without unified visibility over these elements, organizations can perceive AI costs to be rising faster than the corresponding business benefits become visible.
Unified platforms and observability as a remedy
SUSE experts argue that creating a unified AI environment that supports not only model operation but also continuous tracking of costs and performance metrics is crucial. The SUSE AI platform is presented as a solution: an open, enterprise-grade foundation for operating AI services that includes integrated AI-observability capabilities.
These observability features can provide organizations with detailed insights into token consumption, associated costs, and GPU resource performance and utilization. The platform can run on on-premises infrastructure, in the cloud, or in air‑gapped environments, which makes it suitable for organizations requiring strict data control, security, and auditability.
Bottom line
While AI spending has accelerated and some companies already report measurable benefits, many organizations still face challenges in accurately assessing ROI. Improved operational integration, cost transparency, and dedicated observability tools can help firms gain better control over AI expenditures and realize business value in a more predictable way.
