Why Many CEOs Still Aren't Seeing a Return on AI — And What That Can Mean for Finance

Enterprise investment in artificial intelligence (AI) has never been higher. Neither, it turns out, has the gap between that investment and the results it's producing.

PwC's 29th Global CEO Survey helps quantify that gap. Spanning 95 countries, the survey includes responses from more than 4,400 chief executive officers (CEOs). More than half of CEOs surveyed say their company has realized neither increased revenue nor lower costs from AI over the past year. In fact, only about three in 10 report a revenue lift, and roughly a quarter report lower costs. Just one in eight are seeing both. Yet CEOs aren't backing away from AI — they're doubling down on it. But for many of today’s organizations, the financial return still isn't showing up.

The survey's most important finding may explain why. The companies actually converting AI investment into profit aren't the ones running the more pilots or buying the more tools. Instead, the companies realizing profits are those with real AI foundations in place:

• Technology environment built for enterprise-wide integration
• Clearly defined and executable roadmap
• Formal governance and risk processes wrapped around how AI is used

Organizations with those foundations in place are significantly more likely to see meaningful financial returns — and apply AI far more broadly across the business as a result.

In other words, the bottleneck isn't ambition. It's foundation.

Nowhere is that foundation problem more visible than in the Office of Finance. Across the company, Finance sits on top of the more complex, higher-stakes data — general ledger, planning, consolidation, reporting, workforce, and operational data. The data is often spread across enterprise resource planning (ERP) systems, spreadsheets, and point solutions that were often not built to talk to each other.

Layering AI onto that kind of fragmented environment doesn't produce insight. It produces noise:

• Recommendations that can be difficult to trust
• Forecasts that can be difficult to explain
• Outputs that Finance and audit teams can't stand behind when it matters

That's exactly the pattern PwC's survey surfaces at the enterprise level: AI without governed, unified data rarely turns into value Finance leaders can rely on.

To help solve the problem, OneStream's approach — Sensible AI — starts from a different premise than most AI initiatives. OneStream doesn’t bolt AI onto disconnected systems and hope for coherent output. Instead, Sensible AI is built directly into a single, unified platform that already governs how financial data is defined, consolidated, and reported.

That distinction matters because it directly addresses the foundational gaps PwC identifies as the real driver of AI return on investment (ROI):

• A technology environment built for integration — OneStream unifies financial close, consolidation, planning, and reporting on one platform and one data model. That means AI is working from a single source of truth rather than reconciled exports.
• Governance and explainability by design — Each AI-driven output in OneStream is traceable back to governed financial data. That means Finance and audit teams can explain and defend the “why” behind a recommendation, not just the output itself.
• A clear path to scale — Sensible AI lives inside the system of record Finance already trusts. That means the platform can extend from a single use case — say, forecast variance detection — to broader scenario planning and decision support. And Sensible AI does so without requiring a separate integration project every time.

Those drivers underscore the core idea behind Sensible AI: AI that Finance teams can actually stand behind. Why? Because it's accountable to the same data model, controls, and governance, Finance already depends on.

PwC's own prescription for closing the AI ROI gap is direct: Build the foundations first — the integrated technology environment, governance and roadmap. Don’t wait for the “right” moment to start.

According to the PwC 29th Global CEO survey as it is relates to companies that wait, the survey is equally direct: CEOs who hold off on major moves until conditions feel more certain aren't playing it safe. PwC's data shows how the more cautious group grows two points slower and posts profit margins three points lower than peers who moved anyway. The report's own framing for this problem is blunt: The biggest risk isn't dynamism, but denial.

For any organization running SAP, an ERP migration is exactly the kind of multi-year, high-uncertainty program that tempts Finance leaders to defer other priorities else until it's done. That includes deferring the AI foundations PwC says are the real differentiator. Ultimately, deferral is a sequencing decision, not a technology one. Deferral is also a decision that PwC's survey suggests is quietly costing companies more. When “AI-ready foundations” and “ERP transformation” are treated as a single project rather than two, the cost can be high. Finance waits years for AI-ready data, governance, and integration instead of having it all on day one.

An EPM-first strategy can break that dependency. By deploying OneStream's unified platform independent of ERP timing, Finance teams put governed, AI-ready foundations in place now. The platform also supports the timeline PwC's data says matters, not the ERP program's timeline. As a result, Finance can carry that continuity forward regardless of where the SAP migration stands.

Watch OneStream, Microsoft, and PwC for “Modernize Finance Before ERP: Why Leading SAP Organizations Are Choosing an EPM-First Strategy,” a webinar where we walk through why leading SAP organizations are decoupling Finance transformation from ERP timelines and how a SensibleAI approach turns that decision into a competitive advantage.