CFOs shift AI spending focus as infrastructure investment accelerates in 2026

Enterprise CFOs are rebalancing their AI budgets in 2026, emphasising outcome-driven investments and infrastructure growth amid cautious optimism towards AI agents and strategic automation upgrades.

Enterprise spending on generative artificial intelligence (GenAI) is undergoing a significant recalibration as CFOs adopt a more discerning approach to investment in 2026. According to a recent PYMNTS Intelligence survey of 60 enterprises, only 26.7% of CFOs plan to increase their GenAI budgets over the next year, a steep decline from 53.3% a year prior. This shift reflects a move from experimental adoption toward disciplined, outcome-driven deployment amid a changing macroeconomic environment, where interest rates are easing and capital markets are becoming more accessible.

Return on investment (ROI) now dominates AI spending decisions. The survey reveals a clear divide between enterprises experiencing strong financial returns and those seeing limited benefits. Half of the companies reporting very positive ROI intend to boost their GenAI budgets, whereas only 16.7% of firms with negligible ROI share this intention. CFOs are increasingly prioritising financial outcomes such as cycle-time reductions, error minimisation, and working capital improvements over superficial productivity gains. Kevin Akeroyd, CEO of Sovos, commented in an interview with PYMNTS that CFOs are balancing ambition with caution, focusing on how AI can enhance efficiency and reduce regulatory risks, particularly in areas like tax compliance.

Automation maturity also influences budget decisions. Enterprises with advanced automation capabilities are more confident in expanding GenAI investments; 33.3% of CFOs in these firms plan to increase budgets, compared to just 21.4% in less automated organisations. Industry sectors differ too, with 41.7% of technology firms intending to grow their GenAI spending, while services and goods sectors lag behind, at 27.3% and 19.2% respectively. These discrepancies highlight challenges related to integration ease and margin sensitivity across industries.

While enthusiasm for GenAI budgets is moderating, infrastructure investment is accelerating robustly. Data from Menlo Ventures shows enterprise spending on large language model (LLM) APIs more than doubled within six months, from $3.5 billion to $8.4 billion. Anthropic has notably surpassed OpenAI in enterprise deployments, and there is a trend towards favouring closed, proprietary models over open-source alternatives. Despite this infrastructure surge, concerns remain about whether the scale of investment will translate into proportional returns. The Financial Times has cautioned that AI’s financial rewards have yet to meet expectations, warning of overextension risks.

The emergence of agentic AI—autonomous digital agents capable of performing complex tasks—is reshaping CFOs’ strategic focus on AI. Research from Salesforce indicates a dramatic shift over recent years, with conservative AI strategies dropping from 70% in 2020 to just 4% in 2025. More than 60% of CFOs now consider AI agents essential for driving business outcomes beyond immediate financial metrics, appreciating their potential to enhance efficiency and foster long-term revenue growth. Similarly, a study shows that CFOs are allocating roughly a quarter of their AI budgets to AI agents, optimistic about their capacity to reduce costs and increase revenues, with expectations of around a 20% uplift.

However, the adoption of agentic AI remains cautious, with only 14% of enterprises fully deploying these technologies and 34% in the implementation phase. Barriers include cybersecurity, data privacy concerns, regulatory uncertainties, and a lack of internal governance policies, despite widespread positive ROI reported across AI initiatives. Jason Godley, CFO of Xactly, advised in a conversation with PYMNTS that CFOs should balance curiosity with caution, critically evaluating vendor capabilities and potential revenue impacts, while maintaining openness to innovation.

Overall, CFOs display a balanced mindset towards AI investment, testing new technologies judiciously and scaling only where returns are clear and risk manageable. According to PYMNTS reporting, 97% of CFOs trust GenAI for risk management, fraud detection, and forecasting functions. As borrowing costs ease gradually, finance leaders are recalibrating capital allocation with heightened expectations for transparent, measurable outcomes. This pivot marks a maturation in AI adoption, where the focus increasingly lies on embedding proven AI applications and infrastructure rather than indiscriminate expansion.

📌 Reference Map:

• Paragraph 1 – ^[1], ^[7]
• Paragraph 2 – ^[1], ^[4], ^[6]
• Paragraph 3 – ^[1], ^[4]
• Paragraph 4 – ^[1], ^[7]
• Paragraph 5 – ^[1], ^[2], ^[3], ^[4], ^[6]
• Paragraph 6 – ^[1], ^[5], ^[6]
• Paragraph 7 – ^[1], ^[4], ^[6]

Source: Noah Wire Services