Top CFOs warn AI success depends on training employees, not just buying technology

Finance leaders at Fortune’s Emerging CFO webinar said AI is reshaping finance, but companies are underinvesting in the human skills needed to use it effectively, with Deloitte citing 93% of AI spending going to data, technology, and infrastructure versus 7% to people enablement. CFOs from HPE, Prologis, J.M. Smucker, and Moody’s emphasized curiosity, communication, and strategic judgment as automation expands and routine tasks are absorbed by AI. The article is largely a qualitative discussion of workforce transformation rather than a market-moving event.

Analysis

This is less a near-term AI demand shock than a medium-term operating-model shift. The market tends to underappreciate that enterprise AI budgets are still front-loaded into infrastructure, while the real ROI inflection usually comes later, when workflows, incentives, and employee training mature. That means the first beneficiaries are not just the AI platforms, but the workflow layer and implementation layer that can prove measurable labor-arbitrage inside finance organizations.

The second-order effect is that CFO organizations become a distribution channel for AI spend across the enterprise. If finance teams successfully standardize AI-assisted planning, close, and review cycles, they create a template that expands into procurement, HR, and commercial analytics; if they fail, AI initiatives get stuck as capex-heavy science projects with poor adoption. For vendors, that raises the bar: winners will be the ones that reduce time-to-value and governance friction, not those that simply sell more models or compute.

For the named companies, HPE is the cleanest operating leverage story because it can internalize AI use cases while also selling the picks-and-shovels into enterprise transformation, but it also faces the highest execution scrutiny if adoption stalls internally. Workday benefits from the thesis that finance leaders need process redesign, not raw data exhaust, but pricing power depends on proving that its suite shortens cycle times and improves decision quality. Moody’s has a more subtle angle: AI should increase demand for contextual, higher-trust analytics, but only if it avoids becoming another low-differentiation data pipe; otherwise model commoditization compresses product value over 12-24 months.

The contrarian view is that the current consensus may be too focused on tool adoption and not enough on organizational fatigue. In the next 1-2 quarters, companies can announce pilots and training spend, but the real risk is that legacy finance teams treat AI as a layer on top of old processes, creating KPI sprawl without productivity gains. That argues for a selective approach: own vendors tied to workflow redesign and governance, but fade names whose AI story depends on generalized enterprise enthusiasm rather than measurable operating outcomes.

Key Decisions for Investors

• Long WDAY into the next 3-6 months on the thesis that finance workflow redesign is the monetizable layer of AI adoption; favor call spreads over outright stock to capture upside while limiting multiple risk if enterprise spend pauses.
• Long HPE vs short a basket of hardware peers with weaker enterprise software attach rates over 1-2 quarters; HPE has the clearest internal and external proof-point risk/reward if AI becomes an operating expense story rather than only a capex story.
• Buy MCO on 6-12 month horizon only on pullbacks; the market may be underestimating how AI-driven KPI overload increases demand for trusted contextual analytics, but position size should be modest because monetization is slower and more indirect.
• Avoid chasing generic AI infrastructure names here; the article implies adoption, training, and workflow integration are the bottlenecks, so upstream compute beneficiaries may lag until ROI becomes visible in reported margins.
• Pair long WDAY / short a broad software basket over the next 3-6 months to isolate the theme that workflow-enabled AI is winning capital allocation, while generic software names face slower conversion from pilots to billable expansion.