Deloitte Finds Two-Thirds of Firms Increased AI Spending Despite Unclear ROI

Deloitte's 2026 State of AI in the Enterprise survey reveals a troubling paradox: two-thirds of companies increased their AI budgets this year — many by 35% or more — despite only 23% being able to demonstrate clear return on investment from previous AI spending. The survey of 2,800 C-suite executives across 14 industries exposes a market driven more by competitive fear and executive mandate than by proven economic returns.

For CFOs navigating this environment, the data presents a difficult position: underspend on AI and risk competitive disadvantage; overspend without ROI frameworks and face board scrutiny when results don't materialize. The 44% gap between spending growth (67% increasing) and ROI demonstration (23% proving returns) represents the largest investment-confidence disconnect in enterprise technology since the early cloud computing era.

The Spending Numbers

Deloitte's data quantifies the scale of the AI investment surge:

Budget growth:

• 67% of enterprises increased AI spending YoY
• Median budget increase: 35%
• Average enterprise AI budget: $47 million (up from $34 million in 2025)
• Total enterprise AI spending projected: $197 billion globally in 2026

Allocation breakdown:

• Infrastructure and compute: 38% ($18M average)
• AI platform and tools: 24% ($11M)
• Talent and training: 19% ($9M)
• Data preparation and governance: 12% ($6M)
• Change management and deployment: 7% ($3M)

ROI reality:

• 23% demonstrate clear, measurable ROI
• 34% report "some positive indicators but not yet quantifiable"
• 28% acknowledge "unclear returns to date"
• 15% admit "negative ROI when fully accounting for all costs"

The 15% reporting negative ROI is particularly striking — these companies spent millions on AI and can demonstrate that total costs (infrastructure, talent, opportunity cost, disruption) exceeded benefits. Yet many in this cohort are increasing budgets further.

Why Companies Spend Without Proof

Deloitte identifies three primary drivers of spend-despite-uncertainty:

1. Competitive fear (78% cite as top-3 motivation). The most powerful force: executives believe that not investing in AI creates existential competitive risk — even if the specific returns from current AI investments remain unclear. This FOMO-driven spending resembles early internet investment patterns where companies spent heavily on websites without revenue models because "everyone else is doing it."

The fear is not irrational. In financial services specifically, AI-native competitors (neobanks, fintech lenders, robo-advisors) demonstrate 3-5x operational efficiency advantages that will compound over time. The risk of not investing may exceed the risk of investing poorly.

2. CEO and board mandate (62%). Top-down AI directives from boards and CEOs create budget allocations that bypass normal ROI justification processes. When the board says "we need an AI strategy," the CFO funds it regardless of whether prior investments have paid off — because the directive is strategic, not economic.

This creates a dangerous dynamic: AI spending becomes unchallengeable within organizations. Finance teams that would normally demand business cases for $47 million investments accept AI budgets on faith because questioning AI is perceived as being "against innovation."

3. Vendor bundling (54%). Enterprise software vendors (Microsoft, Salesforce, SAP, Oracle) are embedding AI features into existing products and raising prices 15-30%. Companies don't actively choose to increase AI spending — it arrives embedded in renewal contracts. The "AI tax" on existing software represents a significant portion of the budget increase that companies report as "AI investment."

The CFO's Dilemma

Finance leaders face asymmetric risk:

If AI investment succeeds: The company gains competitive advantage, the CFO is praised for foresight, and the investment compounds.

If AI investment fails quietly: Budget continues, no one is fired for spending on AI (just as "no one got fired for buying IBM"), and the company slowly loses competitiveness through misallocation.

If AI investment is cut: The CFO faces immediate criticism for being "anti-innovation," talent retention suffers (AI engineers want to work at AI-investing companies), and any competitor success with AI is blamed on the CFO's caution.

If competitors succeed with AI while you didn't invest: Career-ending. The CEO and board will not accept "we couldn't prove ROI" as an excuse for losing market position.

This asymmetry explains rational overspending: the personal and organizational downside of underspending exceeds the downside of overspending.

What the 23% Who Demonstrate ROI Do Differently

Deloitte identifies patterns among the minority of companies with proven AI returns:

Narrow focus. Successful companies deploy AI against specific, measurable problems (reduce call center volume by 40%, accelerate month-end close by 3 days, decrease fraud losses by $2M) rather than pursuing "AI transformation" broadly.

Baseline obsession. Before deploying AI, they measure current performance exhaustively — establishing clear baselines against which AI impact can be quantified. Without baselines, you can't prove AI worked even if it did.

90-day value sprints. Rather than 12-month implementation cycles, successful companies expect first measurable results within 90 days. If AI hasn't demonstrated value in a quarter, they pivot — not double down.

Outcome ownership. A specific business leader (not IT, not "the AI team") owns each AI initiative's outcomes. When a VP of Operations owns "reduce processing time by 30% using AI," accountability is clear.

Cost attribution rigor. Successful companies track all-in costs: infrastructure, talent, data preparation, business disruption during implementation, and ongoing maintenance — not just the AI platform subscription fee. This honest accounting sometimes reveals that "successful" AI deployments are actually net-negative when fully loaded.

Why This Matters

For CFOs approving AI budgets: Demand the same rigor for AI spending that you'd require for any $47M capital allocation. Establish baselines, define success metrics, set 90-day review gates, and require business-case-owner accountability. The fact that 67% of peers are spending more doesn't justify unjustified spending.

For accounting firms advising clients: The AI spending surge creates advisory opportunities: helping clients build AI ROI measurement frameworks, conducting independent AI investment audits, and benchmarking AI effectiveness against industry data. Firms that can tell a CFO "your AI spending is 2x industry average with below-average returns" provide enormous value.

For board members: Ask: "Of our $47M AI budget, how much has demonstrated returns, and how much is faith-based?" The 23% demonstration rate suggests most boards are approving AI spending without adequate evidence — a governance gap that audit committees should address.

For AI vendors: The 77% of companies that can't demonstrate ROI represent a ticking time bomb of customer dissatisfaction. When boards eventually demand evidence, vendors without clear value attribution will face budget cuts and churn. Building ROI measurement into your product (not just features) is a competitive imperative.

For the finance profession: AI accounting is becoming a specialty. Capitalizing AI development costs (ASC 350-40 vs. expense), measuring AI asset useful lives, assessing AI investment impairment, and building AI-specific internal controls represent emerging practice areas where demand is growing faster than expertise.

The key takeaway: Deloitte's finding that 67% of firms are increasing AI budgets while only 23% can prove ROI reveals a $150 billion annual faith-based spending pattern — CFOs who impose rigorous measurement frameworks now will either justify their AI investments or reallocate capital to proven use cases before boards demand accountability.