Flexera report shows organisations use AI but lack skills

All organisations in Flexera's latest cloud survey now use public cloud services for generative AI, while 73% report lacking the resources or expertise to manage their cloud environments effectively.

The findings point to a widening gap between AI uptake and companies' ability to control the cloud systems behind it. In the survey, 45% of respondents said they use generative AI public cloud services extensively, suggesting adoption has moved beyond pilot projects in many businesses.

Cost emerged as the clearest operational concern. Some 85% of organisations surveyed said managing cloud spend is their top cloud challenge, while 30% said cost unpredictability is a major issue when scaling AI workloads.

As companies push AI tools into daily work across departments, governance, budgeting and technical oversight have often struggled to keep pace. That shift has increased pressure on IT teams to monitor usage, allocate spending and understand how quickly demand for cloud computing can rise once AI services spread across a business.

Skills gap

Flexera based the findings on a global survey of 753 technical professionals and executive leaders conducted in winter 2025. Respondents included cloud decision-makers and users across a range of industries, organisation sizes and job functions.

The survey suggests cloud management problems are no longer limited to early-stage adopters. If every organisation surveyed is now using generative AI cloud services in some form, the question is no longer whether companies will adopt these tools, but how they will manage them without letting costs and complexity grow unchecked.

That is especially relevant for businesses trying to move from isolated AI use cases to broader deployment. Generative AI services can quickly increase demand for computing resources, particularly when multiple teams use them at the same time for different tasks.

This can make budgeting harder. Unlike more predictable software spending, AI-related cloud use can fluctuate with employee demand, experimentation and the volume of data being processed, leaving finance and technology leaders with less certainty over how costs will develop.

Marlon Oliver, Senior Vice President, EMEA, at Flexera, said: "Many organisations are moving quickly to scale AI across the business, but the operational reality is that most teams are still struggling with cloud cost, complexity and visibility. AI introduces a completely different level of demand on cloud environments, particularly once usage starts growing across multiple teams and workloads. The risk for organisations is that they focus heavily on AI adoption without investing at the same pace in the expertise needed to manage these environments effectively."

Rising pressure

The research adds to a broader debate over whether rapid workplace AI adoption is translating into clear business returns. Companies have expanded access to AI assistants and models across functions ranging from software development to customer service and internal administration, but many are still working out how to measure productivity gains against rising infrastructure bills.

For cloud providers and software vendors, that expansion has created fresh demand for AI-related services. For customers, it has introduced a more complex operating model, with spending tied not just to software licences but also to consumption of processing, storage and model usage.

Flexera's data suggests many organisations have accepted AI cloud services as part of normal operations even though they remain unsure they have the internal skills to manage them well. The combination of universal usage in the survey and high concern over expertise indicates adoption is being driven as much by business expectation as by technical readiness.

That mismatch may become more visible as companies try to tighten spending discipline. Businesses that once treated AI trials as a modest innovation cost may now face broader questions about procurement controls, chargeback models, cloud oversight and who is accountable for monitoring usage across departments.

The 30% figure for cost unpredictability points to a distinct issue in AI deployment. Even where organisations are willing to spend on cloud services, unexpected swings in consumption can make budgets harder to forecast and make it more difficult to judge whether an AI project is delivering value in proportion to its cost.

Oliver said: "The risk for organisations is that they focus heavily on AI adoption without investing at the same pace in the expertise needed to manage these environments effectively."