Flexera’s 2026 IT Priorities Report surveyed 834 IT leaders across the U.S., UK, Germany and Australia, and the year-on-year shift is unambiguous.
This is no longer a story about whether organisations will adopt AI. It’s a story about whether they can prove it’s working, govern what it costs, and keep the rest of the estate from becoming collateral damage in the process.
That is precisely the mandate our DAPA and Sprint Zero frameworks were built around, and this year’s numbers make the case more clearly than last year’s did.
The proof-first mandate just became explicit
Last year, we noted that 85% of IT leaders felt prepared to leverage generative AI while 90% admitted their people’s skills needed to evolve.
This year’s report sharpens that tension considerably. 93% of IT leaders say they’re regularly asked to innovate faster and demonstrate ROI in the same breath, yet only 19% say demonstrating AI usage and effectiveness is actually a top priority for the year ahead.
The report cites MIT research putting a hard number on the consequence: despite $30-40 billion in enterprise generative AI investment, 95% of AI projects are returning zero. That single statistic is the clearest external validation we’ve seen for why Phase Zero exists. Readiness assessment before scale-up isn’t caution for its own sake. It’s the difference between joining the 5% and joining the 95%.
Notably, the raw enthusiasm number for AI integration actually fell this year, from 46% of leaders naming it their top priority in 2025 to 33% in 2026, even as AI held the top spot for a third consecutive year. Flexera’s own analysis attributes part of this to a new response option (demonstrating AI effectiveness, which pulled 19% on its own), but the direction matters more than the mechanism: the conversation is visibly shifting from “are we integrating AI” to “can we prove it worked.” Brian Shannon, Flexera’s CTO, frames the discipline required plainly in the report:
“You have to think ‘AI-native.’ AI is going to be a part of everything we do.”
That’s an operating model, not a feature.
It’s the same principle behind DAPA: AI integration has to be designed into the platform and governance layer from day one, not retrofitted once the pilot succeeds.
AI overspend just became the single largest cost concern in the survey
This is the most consequential shift in the entire report for anyone running enterprise cost governance. Last year, security tools topped the overspend list at 31%.
This year, security tools overspend dropped six points to 25% and fell to third place. AI applications took the top spot outright, at 36%, up from an equivalent 23% for “emerging applications” the year before, a 13-point jump in twelve months.
At the same time, 80% of IT leaders report increased AI spending and 62% report increased FinOps investment, up sharply from 48% in 2024. The instinct to govern is strengthening. It simply hasn’t caught up with the speed of AI spend growth yet. As the report puts it directly:
“The AI gold rush may be outpacing governance.”
This is the exact gap our work with Vodafone Portugal was built to close at scale, where we now govern more than 2,900 pipelines precisely because uncontrolled growth, whether in pipelines or AI workloads, produces cost and risk exposure long before anyone notices it on a dashboard.
The report also flags a genuinely new pressure point worth calling out to any CFO reading this: 27% of IT leaders now cite changing pricing models, usage-based, workload-based, outcome-based, as a top spend optimisation challenge, a category that essentially didn’t exist as a distinct concern twelve months ago. Our operational discipline applies directly here, translating volatile, consumption-based AI pricing into forecastable, governable cost structures rather than treating it as an unmanageable variable.
Visibility gaps are getting worse, not better
Eighty-five percent of IT leaders now say gaps in IT visibility pose a risk to their organisation, a six-point increase on last year. Fifty-eight percent report specific problems from unsanctioned or shadow SaaS usage, and 70% believe business units are procuring more cloud and SaaS than IT is even aware of, up from 68%. Nearly nine in ten (87%) are now consolidating point tools into unified platforms, up from 84% in 2024, which validates the platform-first approach we’ve taken with every DAPA engagement rather than layering yet another disconnected tool onto an already fragmented stack.
One barrier is worth flagging specifically because it’s moving in the wrong direction: the share of leaders citing “not having the right tools for visibility” as a barrier rose from 16% to 21% this year. That’s not a governance failure so much as a tooling gap, and it’s precisely the kind of structural problem a phased, platform-led approach is designed to close before it compounds further.
The vendor landscape is consolidating around AI leaders
Microsoft held its top vendor position for a fourth consecutive year, rising from 63% to 66%.
Google jumped ten points to 59%. OpenAI held steady at 37% but now stands alone in fourth place, no longer tied with Oracle as it was last year. Oracle itself fell ten points to 27%, and the wider second tier, SAP, Salesforce, Adobe, ServiceNow, Autodesk, Broadcom, all dropped by roughly nine to eleven points apiece.
For enterprise clients, this consolidation has real implications for platform strategy and vendor risk. It reinforces why Teraflow’s advisory work stays vendor-agnostic at the architecture level: when the vendor landscape moves this fast, in one year, the client’s governance framework needs to outlast any single partner’s market position.
What this means for SMBs, not just enterprise clients
The same three pressures apply at SMB scale, only with less internal capacity to absorb them.
AI spend is growing faster than governance capability (80% of leaders report AI spend increases, only 19% plan to prove ROI). Cloud and SaaS visibility is degrading rather than improving (six points worse in twelve months). And a genuinely new cost variable, AI pricing model complexity, has entered the picture with almost no mature tooling built around it yet.
This is exactly why Sprint Zero exists as a starting point regardless of company size: establish readiness and governance before committing budget to scale. It’s why DAPA is built to make platform decisions durable against a vendor landscape that reshuffled dramatically in a single year. And it’s why FloJo keeps operational discipline attached to every AI workflow, so that the 95% failure rate MIT describes becomes a statistic our clients read about rather than live through.
The report’s own framing captures the moment precisely:
“Organisations that succeed will treat AI as a discipline, complete with accountability.”
That’s not a new idea to us. It’s the idea DAPA and Sprint Zero were built on. What’s changed in twelve months is that the data now makes the cost of ignoring it impossible to miss.
Source: Flexera 2026 IT Priorities Report, based on a survey of 834 IT leaders across the U.S., UK, Germany and Australia. Comparisons are to the Flexera 2025 IT Priorities Report (N=800).
